ECONOMETRIC MODEL OF SKI RESORT REAL ESTATE
IN NEW ENGLAND
by
William Daniel Gause
Master of Engineering
Cornell University, 1988
Bachelor of Science
Cornell University, 1987
SUBMITTED TO THE DEPARTMENT OF ARCHITECTURE
IN PARTIAL FULFILLMENT OF THE REQUIREMENTS OF THE DEGREE
MASTER OF SCIENCE IN REAL ESTATE DEVELOPMENT AT THE
MASSACHUSETTS INSTITUTE OF TECHNOLOGY
SEPTEMBER, 1993
@William Daniel Gause 1993
All rights reserved
The Author hereby grants to M.I.T.
permission to reproduce and to distribute publicly copies
of this thesis document in whole or in part.
Signature of the author_________________
William D. Gause
Department of Architecture
July 31, 1993
Certified by
William C. Wheaton
Professor of Economics
Thesis Supervisor
Accepted by
William C. Wheaton
Chairman
Interdepartmental Degree Program in Real Estate Development
RatctIl
MASSACHUSETTS INSTITUTE
OF TFCANOLQGy
OCT 04 1993
LIBRAiltib
ECONOMETRIC MODEL OF SKI RESORT REAL ESTATE
IN NEW ENGLAND
by
WILLIAM DANIEL GAUSE
Submitted to the Department of Architecture
on July 31, 1993 in partial fulfillment of the
requirements for the Degree of Master of Science in
Real Estate Development
This paper is an economic analysis of residential ski resort real estate in New
England. In an attempt to understand what economic, geographic, demographic, and
climatological factors influence the price of second homes at ski resorts, predictive models
were generated for skier visits, condominium rental rates, hotel rental rates, condominium
prices and construction completion rates. These models are based on data collected at a
single New England ski resort recognizing that most large New England ski resorts are
similar and yet are characteristically different from large Western resorts. This data was
collected from historical records dating back twenty-five years, covering several booms
and recessions.
The primary purpose of the study was to determine what general factors influence
second home prices in New England so that developers and investors would be more
knowledgeable in making decisions about when and where to build or invest. Specific
models were created for the five condominium projects whose data was used, and these
models can be used to predict future prices and rates at these specific projects. The
trends, however, should hold true for other New England resorts. The skier visit equation
should hold true for all large, southern Vermont ski areas since this model was based on
data for the entire state.
The results of the study indicate that skier visits are primarily a function of regional
employment and snowfall, with increasing significance associated with employment and
decreasing significance associated with snowfall. Condominium rental rates are largely a
function of skier visits, condominium stock, total regional employment and the previous
year's rental rate. Hotel rates are primarily a function of skier visits and the previous year's
rate. Interestingly enough, condo rates are more a function of the previous year's skier
visits while hotel rates are better correlated with the current year's skier visits indicating
that hotels do a better job of estimating future visits while condominiums estimate this
year's demand based on last year's demand. Prices are largely determined according to a
consumer's model as opposed to an investor's model. Prices are a function of
employment, skier visits, stock and previous prices rather than a function of rent, interest
rates and inflation. This is a useful characterization of the second home market.
Condominium completions are primarily a function of the price, rent, stock and
employment.
Based on projected employment figures for the next decade and a half, the paper
uses the models to make forecasts of the movement in the ski real estate market in New
England. With forecasts of slow but positive job growth in New England and the midAtlantic states throughout the year 2010, the models indicate a continued slight drop in
skier visits and rents over the next couple of years followed by steady, gradual increases
through 2010 as employment edges up. Prices have reached their bottom and should
climb gradually over the next decade and a half without reaching their mid-eighties highs
(in real terms) over this period. Completions should experience a brief surge over the next
two years before tapering off for a couple years and then beginning a slow steady climb,
assuming there are no physical or legal impediments to growth. The degree to which
these trends occur is, however, entirely dependent upon the level of employment growth in
the region. Employment is the common thread which binds the second home market
together.
Thesis Supervisor:
William C. Wheaton
Title:
Professor of Economics
TABLE OF CONTENTS
Abstract.......................................................................................................
2
Introduction................................................................................................
5
Factors Affecting the Value of Ski Resort Real Estate...............................
11
M ethodology of Study .................................................................................
22
Skier M arket..............................................................................................
38
Rental M arket............................................................................................
45
Asset M arket..............................................................................................
56
Forecasting for Ski Resort Real Estate ......................................................
76
Conclusions....................................................................................................
110
Appendix........................................................................................................
117
Bibliography ..................................................................................................
131
%,,..........
xx x
.......
............
..............................
...................
.........
............
...
...
.
. ....
Purpose of Study
The purpose of this thesis is to track and analyze the history of ski resort real
estate prices and rental rates over time and to index them to economic, geographic, and
atmospheric factors. The intent is to be able to use these models to forecast future skier
visits, rental rates, prices and completions. The study was conducted at a single New
England ski resort (Killington/Pico Vermont) for the purpose of determining trends in the
industry. The specific models and data can only be applied to the Killington area, but the
trends can be used to draw parallels to other New England ski resorts and possibly for
other New England recreational real estate in general. If it is possible to estimate a surge
or drop in real estate prices by following trends in the regional economy this could be
invaluable to a real estate owner or investor. The purpose of the study is to analyze and
predict trends in the second home market for recreational developments rather than
specifics from one year to the next. If someone is interested in specific models, they
should follow this format to conduct a similar study at the specific resort of interest.
Description of New England Ski Resorts
James Branch of Sno-Engineering Inc. has classified ski resorts into four different
types according to their size and characteristics'. Type I resorts are large world-class
destination resorts that have many amenities and are often built around a city or town.
They often have a lot of foreign investment and well-developed real estate. Examples of
such resorts include Aspen, Snowmass and Vail. Type II resorts are similar in size to
Type I resorts, but they often don't have as many amenities, aren't built around a thriving
IPatrick Philips, Developing with RecreationalAmenities: Golf Tennis, Skiing andMarinas
(Washington D.C.: Urban Land Institute, 1986), p.
resort village, and don't attract the same amount of foreign investment. Examples include
most large New England resorts such as Killington, Sunday River, Mount Snow and
Sugarloaf Type III resorts are smaller than Type II resorts and don't have the same
amount of associated real estate. Examples would include smaller New England resorts
such as Bromley, Mad River Glen, and Shawnee Peak. Finally, Type IV resorts are small
local ski areas that are geared towards local skiers and do not have much in the way of
real estate development.
Type I resorts are destination resorts that attract visitors from all over the world
for extended stays of one week or longer. Branch argues that people investing in real
estate at these resorts are older and more interested in the private use of the real estate.
They are less concerned with rental income using the property as an investment tool.
Type II resorts are generally more regional. While they draw people for weeklong visits, they often do not have commercial airports and service. They are more
dependent on vehicular traffic and rely on large metropolitan areas within driving distance.
As such they, they are often the target of weekend visits. According to Branch, people
investing in real estate at Type II resorts are more likely to rely on the rental income from
the property. They are often younger and may prefer to ski at a variety of different
mountains rather than always skiing at the same place.
Type III resorts are often considered weekend ski areas with primarily local
patronage during the week. Type IV resorts are often open only at night and on the
weekend, catering only to local skiers.
This study looks at Type II resorts in New England. As mentioned above, Type II
destination resorts are often dependent upon people driving to the area. Assuming that
most people don't want to spend more than about five hours in a car, New England resorts
are primarily dependent upon the Boston and New York metropolitan regions. They are
obviously exceptions to this rule, but for the purposes of this study it will be assumed that
Massachusetts, Connecticut, New York and New Jersey are the primary markets for skiers
who would be interested in investing in ski resort real estate at Killington. Although there
are many skiers from Vermont, New Hampshire and Maine, it is assumed that they are not
the primary investors in Killington's resort real estate, since they either already live in the
area, or are closer to another mountain.
New England ski resorts are characteristically unique. They are often located near
small, quaint New England villages. They are usually situated in small mountain ranges
(when compared to the Rocky Mountains) and have skiing trails lined with trees (not
cliffs, crevasses and couloirs). They are at a lower elevation than western resorts which
means more modest snowfall amounts and a dependence on artificial snowmaking. New
England ski resorts generally have one season - winter, although some areas are trying to
change that. The character of New England resorts is distinctly different from Western or
European resorts and hence has different market appeal. The results of this study are
intended to be applicable to Type II New England resorts only. An analysis of Western
resorts would have to consider a host of different factors.
Types of Ski Resort Real Estate
Residential ski resort real estate consists primarily of single family detached homes
i.e. "chalets", and condominium units. Up until the late seventies and early eighties, single
family housing was the most common form of ownership, with relatively few
condominium projects. Since then, however, the popularity of condominiums has
skyrocketed. While condominiums do not offer the solitude and privacy of a single family
home, they have many features which make them very attractive to second home buyers.
The primary attractions are their lower price, ease of renting, and lack of maintenance.
Many people with ski resort homes rent them out when not in use to help cover the cost of
ownership. Condominiums have similar features and are often listed with a single broker
to facilitate renting. Finally, with all the upkeep of a primary home, many people prefer
not to deal with twice as many headaches and opt for maintenance free condominiums.
Another increasingly common form of ownership is the timeshare. These are much
more affordable than buying a condominium outright and allow the occupant to pay only
for the time they use the unit. It offers much more flexibility than owning a single unit
since shares may be traded and the person may easily buy shares at more than one resort,
giving them variety. It also eliminates an owner's reliance on renting the unit. The
popularity of timeshares may increase in times of financial belt tightening due to their
lower cost. The primary disadvantage is their lack of flexibility and their multiple
ownership. Vacation planning has to be coordinated well in advance. Many people in the
field expect to see a surge in the demand for timeshares. 2
The other primary type of ski resort real estate is commercial. This includes
hotels, restaurants, shops, and other places of business. This study will consider the
fluctuation of hotel rents as a measure of the demand for accommodations, however, it
will be focused on residential real estate and in particular it will be concentrated on
condominiums. It is assumed that the demand for commercial real estate roughly follows
the demand for residential real estate. The more people there are living in the area, the
greater the commercial demand will be.
The Market for Ski Resort Real Estate
Ski real estate can be thought of in two types of markets--the investment market
and the consumer market. Some people buy resort homes for the investment potential
while others buy them to use while skiing or relaxing in the country. As will be discussed
later, this leads to two different market models with different input factors.
Statistics indicate that the bulk of the skier market is young, with the average skier
between 18 and 35.3 The majority of the second home buyer market is over 35, however,
2"Resorts
3Gregg
and Recreation", Urban Land Special Trends Issue (March 1991), pp. 38-39.
T. Logan and Ann E.Day, "Ski Resort Real Estate Development", UrbanLand
(December 1989), p.37.
and typically an empty-nester or mature family aged 40 to 55.4 A 1990 study found that
while people aged 35 to 64 accounted for 50% of all American householders, they
accounted for 70% of the recreational property owners in the country.5 According to
Logan and Day, today's typical second home buyer wants to retain private use, are less
concerned with rental income, want to use the unit in the off-season as well as the peakseason, and frequently use the property as a means of bringing together extended families.
These are people who can afford to purchase a second home after owning a primary
residence. They are looking for more and more amenities and are primarily interested in
large condominiums and townhouses. 6
Given the profile of the typical second home buyer and the demographic
composition of the United States, this country may be on the edge of a surge in the
demand for second homes in the coming decade as the "baby-boomers" reach this age
group. It is important to note, however, that the "baby-boomers" are aging out of the
range of the average skier. As these people age and look to buy second homes they are
going to be looking for more amenities than just skiing. It will be increasingly important
to provide year-round amenities to attract this potentially huge market.
Leanne Lachman, of Schroder Real Estate Associates in New York, predicts a
doubling of the demand for second homes over the next decade as the nations first
generation of dual-income couples reaches middle age with discretionary time and money.7
This demand should occur not only in ski homes, but primarily in year round second
homes. To cash in on this market, ski resorts will have to offer year round amenities.
4
Ibid.
5Judith
Waldrop, "Who Owns Recreational Property?", American Demographics(May 1991),
6Logan
and Day, "Ski Resort Real Estate Development", p. 37.
p.49.
7 Susan Bradford,
"Second Homes", Builder (February 1991), p. 79.
Another factor to consider is the changing lifestyles of Americans. As some
people have less leisure time they want to spend it more efficiently. This often means
taking mini-vacations instead of the traditional one or two week vacations. Given these
trends, the demand for second homes located within several hours of the primary home
should surge. 8 This in turn may spell good news for New England ski resorts, as fewer
people from New York and Boston look to spend a week or two out West and instead
spend more weekends at New England resorts.
Literature Review
The bulk of the literature and research dealing with second home and resort
developments is qualitative rather than quantitative. There are many articles speculating
that a boom in the second home market is about to occur due to the aging "baby-boomers"
and the traditional second home buyer's profile. The problem, however, is that most of the
literature is based on opinion and reasoning rather than empirical studies. To the best of
the author's knowledge, an econometric model of ski resort real estate has never been
conducted that looked at the sales and rental history of an established group of real estate
and tried to fit that data to economic, geographic, atmospheric and demographic factors
over time. If such a study has been conducted, it was not published in a manner that made
the results widely accessible. The literature that is accessible has to do with market
potential and expected future opportunities in a very general sense. This paper will
attempt to broaden the scope of the literature dealing with the subject.
8Diane
p. 32.
R. Suchman, "Opportunities for Recreational Developers", Urban Land (February 1991),
~.0
Overview
Ski resort real estate values are a function of many different variables including
economic, atmospheric, geographic and political influences. The issue at hand is whether
it is possible to develop an econometric model to accurately forecast what these values
will be over time as the input variables change. In developing such a model, some things
are easily quantified while others are obviously not. The following discussion is intended
to highlight some of the more prominent factors influencing real estate values. Following
this discussion, is a description of the specific model developed and the factors considered.
National Economy
The national economy clearly has some effect on the value of recreational real
estate, however, that effect may be secondary and not direct. The value of ski resort real
estate is a function of the economy that fuels the resort. In the case of a national
destination resort such as a "Class I" Western resort, e.g. Aspen or Vail, the visitors are
drawn from across the country and hence the national economy may have some direct
effect upon skier visits and hence real estate prices. On the other hand, a New England
resort draws visitors primarily from New England and the mid-Atlantic states. The
regional economy is what drives the number of skiers visits and the national economy may
have only a secondary effect upon the resort prices by affecting the regional economy.
As will be shown by the model, national mortgage rates have only a limited effect
on the value of ski resort real estate.
Regional Economy
As mentioned, the number of skier visits to a resort and the price of real estate at a
resort is a function of the employment and the economy in the region comprising the skier
market. In the case of weekend destination resorts such as those found in New England,
this means that the regional economy has a direct effect upon the value of the real estate.
As the employment in New York City and Boston drops, this signals a drop in the regional
economy. People are less willing to spend money as their disposable income is reduced
and are less likely to go skiing. As the number of skiers drops, so does the rent and the
value of the resort real estate. When employment increases in these regions, we see the
opposite effect.
Local Economy
The same logic as mentioned above holds true for the local economy. In the case
of a "Class IV" local ski area, the ski area is dependent upon the local residents for
revenue. If the local economy falters, the ski resort will falter as people's discretionary
income is reduced. In the case of these local ski areas, however, residential resort real
estate is not a factor. The skiers already live in the area and are not going to buy a second
home on the mountain.
On the other hand, when we consider a Class II weekend resort, the local economy
does not have much of an effect upon the ski area. Rather the local economy is dependent
upon the regional economy. As the regional economy falters, fewer people make trips to
the ski area which means lower real estate prices and fewer jobs needed in the local
economy. This cycle is even more pronounced when we consider a Class I resort. The
national economy affects skier visits to the resort which in turn affects the regional
economy and the local economy. This inter-dependence clearly can have both positive and
negative effects for the sub-economies.
National Real Estate Cycle
The national real estate cycle is difficult to quantify and analyze. Real estate
markets are local and are a function of local supply and demand. The only factors of the
national real estate cycle that may have some effect on ski resort prices are the average
mortgage rate and the consumer price index, which combine to yield the real mortgage
rate. The rationale would lead one to expect that as the mortgage rate is lowered, more
people can afford to buy real estate and the increased demand causes upward pressure on
prices. The econometric model was unable to prove this with any significance, and in fact
mortgage rates were more often found to have the reverse effect. In general, mortgage
rates were found to have limited effect upon real estate prices.
Regional Real Estate Cycle
As mentioned above, real estate markets are local and are a function of local
supply and demand. Regional real estate is a function of the stock and demand in that
region. If regional prices fall due to increased stock, this should not have a direct effect
upon resort prices. If on the other hand, regional prices fall due to lack of demand
resulting from lack of employment or income, this should affect resort prices. There is,
therefore, no direct correlation between the regional real estate cycle and the prices of ski
resort real estate. They are dependent upon different factors as well as some common
factors. This study did not consider the effect of regional real estate cycles on the value of
resort real estate. A more comprehensive study might include this data if it could be
obtained to see if the two cycles have any correlation.
Local Real Estate Cycle
Resort real estate is clearly a function of the local second home market, which is a
function of the local real estate cycle. Supply and demand drive the local market. As
supply in the local area increases, prices will drop unless they are offset by commensurate
increases in demand. This demand can be brought on by primary or secondary home
buyers since they could both be competing in the same market. It is, however, more
probable that resort real estate buyers will be competing in the second home market, since
they will have a higher utility associated with a mountainside location than a primary home
buyer. They will also be more affluent people transferring wealth from a region with a
higher cost of living (a dollar will not be worth as much to them as it will to a local
resident).
Stock of Ski Resort Real Estate
From basic economics, it follows that real estate prices are a function of the total
stock. The greater the supply of potential choices, the lower the prices of those choices
will be assuming constant demand. As demand increases, however, an increasing supply
will only mean reduced prices if the supply grows faster than the demand. If the supply
keeps pace with the demand, real prices should remain constant (all other things equal). In
predicting real estate prices, it is just as important to be able to predict stock as it is to be
able to predict demand.
Number of Skier Visits
Real estate rent and prices are a function of skier demand. The more people there
are skiing, the more people are going to need accommodations. As will be shown with the
model, skier visits are more important for predicting rental rates than they are for
predicting sale prices. A willingness to ski does not necessarily create a willingness to
buy.
Ski Resort Characteristics
The individuality of a resort has an obvious effect on the value of real estate. Each
resort has its own individual characteristics that add charm to the resort and draw a certain
crowd of people. Some amenities are quantifiable and others are not. The addition of a
golf course may have a measurable increase in the value of the real estate (the increased
value should more than offset the cost of the golf course) by adding a year-round
attraction to the resort and hence increasing the utility.
On the other hand, some amenities effect on real estate can not be quantified. The
ambiance of a resort can either draw people or turn them away. It is important for a resort
to keep in mind the reason for its success. The owner of Sunday River Ski Resort in
Bethel, Maine attributes the phenomenal growth and success of his resort to maintaining
an awareness that visitors are there to ski. They demand the best ski terrain, snow
conditions and lift access, and Sunday River has responded by continually putting money
back into the skiing to make it top notch.9 In the mid-eighties, Sugarloaf/USA Ski Resort
in Kingfield, Maine lost sight of this principle and paid the price. They focused more on
real estate than on skiing. The lack of ski related capital improvements led to reduced
skier visits and hence reduced demand for real estate. Sugarloaf/USA became over built
and was forced to enter Chapter 11 bankruptcy protection. They have since reemerged
from protection and have been putting money back into the skiing since.10
This model assumes that a ski area will make the necessary capital improvements
to maintain a flow of skier traffic. As the volume of skiers increases, the ski area has to
expand to accommodate the increase. As the character of the skiers changes, the ski area
has to change to accommodate the new skier. The character of the resort plays an
important role in determining the value of the real estate and should not be
underestimated.
9 Author's
interview with Leslie Otten, Owner of Sunday River Ski Resort, Bethel, Maine,
January 28, 1993.
10Author's
January 30, 1993.
interview with Warren Cook, Owner of Sugarloaf/USA Ski Resort, Kingfield, Maine,
Residential Site/Unit Characteristics
The characteristics of the individual site have an obvious bearing on the value of
the real estate. While this paper will try to predict future values of real estate prices, it can
only do so accurately for the specific resort and for the specific properties analyzed. The
trends, however, can be used to predict the value of new units considering the individual
characteristics of each. As will be shown, for example, the more floors a unit has the
higher the price will be for the same square footage. Many characteristics, however, are
unique and can not be quantified. It is important to bear this in mind when selecting a site
and designing units.
Environmental/Land Use Regulations
Clearly the more restrictive the regulations in an area, the higher the price of real
estate will be. The stock of real estate may be held artificially low by constraints on the
free market such as zoning restrictions or a long and costly permitting and approval
process. These factors are not considered in this model. All the units are located within
the same town and presumably were subject to the same regulations.
Taxes
As with all real estate, taxes help determine the market value. The higher the
taxes, the less someone is going to be willing to pay. The advantage of second home ski
resort real estate is that the property taxes are often low due to the nature of the residents.
The residents are not year-round and hence do not require the services of primary
residents such as education. From a primary residents perspective, this situation is ideal.
They have outside money from part-time residents who aren't requiring services and aren't
filling local jobs.
Federal income tax treatment of second homes is the same as that for primary
homes, namely that mortgage interest is deductible against Federal income. Although
there have been attempts to modify this tax policy over the years, it has been in effect for
the period of this study and hence should not affect the results. A change in this policy,
however, would certainly have repercussions in the second home market. If the allowance
is reduced or eliminated it will result in a reduction of resort real estate prices across the
country as properties effective costs increase.
Given that all these properties were in the same town and have experienced
roughly the same tax rate over the years and that the Federal tax policy has been constant,
taxes have not been included in this study although they certainly have an effect on the
value of real estate.
Another important aspect of Federal income taxes is the deduction allowed for
rental income properties. Prior to the Tax Reform Act of 1986, owners of rental property
were able to depreciate the property and deduct the mortgage interest against their
personal income. When this was discontinued in 1986, the attractiveness of investment
properties was reduced. One would expect therefore that after 1986 the investment model
of prices would be less accurate. In fact, however, the results indicate that prior to 1986
the investment model had a worse fit than the investment model for the entire period of
the study. The effect of TRA '86 has been left out of the study as well.
Financing
One would think that financing would be a major concern in determining real
estate prices. As interest rates go down, people can afford to spend more for the same
monthly payment. As will be shown, however, mortgage rates have a limited effect on
prices. In fact, the models indicate that there may even be a positive correlation between
mortgage rates and prices, which would be counter intuitive. This would suggest that as
mortgage rates go up, prices go up.
Description of Model
Considering these factors, this paper assumes a flow diagram for determining real
estate values as shown in Figures 1 and 2 on the following pages. It is assumed that
employment and snowfall are the two primary exogenous variables driving the value of ski
resort real estate. These two variables combine to determine the annual skier visits to a
region."1 These annual skier visits are used as a gauge of demand. The more people there
are skiing, the higher the demand for real estate will be. The stock of condominium units
is obviously a measure of the supply. The more units there are, the more units there will
be for sale. The demand (visits) and supply (stock) combine to determine the market rate
for condominium rent.
The condominium price index is then determined according to one of two basic
models, the consumer model and the investment model. The consumer model assumes
that the primary buyers of resort real estate are consumers who want to use the condo or
house as a second home for their own enjoyment. They are not driven by the investment
potential, but rather by their own utility. The investment model on the other hand assumes
that buyers are primarily interested in property for the investment value. They would
value real estate based on the potential cash flow rather than the pleasure derived from the
use of the property.
In the consumer model, the price index is a function of the regional employment,
the condominium stock, and the skier visits. As regional employment and visits increase,
demand increases. As completions occur, they lead to increases in stock which must be
met with increases in demand in order to prevent prices from falling. In the investment
model, the price index is a function of the condominium rent, the mortgage rates and the
consumer price index (a measure of inflation). Increased rent should lead to increased
prices due to the increased value. Lower mortgage rates should result in higher prices
since more people will want to buy when rates are low, driving up the price. High
IAlthough New England ski resorts usually have snow cover regardless of the natural snowfall
due to snowmaking capabilities, natural snow plays into the psyche of skiers. Skiers subconsciously think
that unless there is snow on the ground, the skiing can't be that good.
inflation should result in higher prices since more people will be investing in real estate as
a hedge against inflation.
Finally, the condominium construction is a function of the condominium price
index and the condominium rent index. As the prices and rents go up, developers will be
more likely to build new units in order to capitalize on the higher rates.
The rent index, price index and completions are all interdependent, endogenous
variables. In addition to each other, they are dependent upon skier visits and the few
exogenous variables in the model: employment, snowfall, mortgage rates and inflation.
The hotel rent is primarily a function of the skier visits. The stock of hotel rooms
may have some significance in determining hotel room rates, but was unavailable and was
not used in the model.
Flow Diagram of Consumer Price Model
Figure 1
Flow Diagram of Investment Price Model
Figure 2
.I
... ......
:::::.::.:~
...........
*
**
Assumptions
The primary assumptions of the model are as follows: 1.) We can develop
econometric models that can accurately predict the number of skier visits, condominium
rent, hotel rent, condominium price, and condominium completions by inputting forecasts
of employment, mortgage rates, inflation and snowfall. 2.) By studying one New England
resort that has an established history, we will be able to predict trends at other New
England ski resorts. 3.) The resort that has been selected (Killington/Pico) is driven by
the conditions in Vermont and New England and does not itself create or drive the data.
4.) We can use the results to draw conclusions about other recreational development in
New England.
Characteristics
Study Area
The area chosen for study was the Killington/Pico resort region located in
Sherburne, Vermont. The reason for choosing Killington/Pico is that they are two
of the oldest ski resorts in New England and have a long history of development.
They were the site of some of the first ski condominiums in New England in the
late sixties and early seventies. The areas have prospered over the years and are
still favorites among many skiers.
The other advantage of the Killington/Pico resorts is that they are both
located in the same small town of Sherburne, Vermont and between the two of
them comprise the majority of the town. They are approximately 2,000 year-round
residents and approximately 15,000 part-time residents.12 Every condominium
complex in the town except for one is either on the Killington access road or at the
base of Pico Mountain. This simplified the data collection process.
Time Frame
In order to develop a good model of what drives real estate prices, it was
necessary to consider a long time period. For this model, a time period of 24 years
was used going back to 1969. By doing so it was possible to see the ups and
downs in the overall economy over the past two decades including booms and
recessions and track the effects.
Type of Real Estate
In order to develop an accurate model, it was necessary to look at the
sales, resales and rents of the same units over the period of the study. By doing
so, I was comparing apples to apples so to speak, and was eliminating the external
effects of variation between a variety of properties. Since condominiums are
usually built with identical units, it was possible to have a sale of only a couple
units in each project in each year in order to get a sense of the price index. This
would have been much more difficult to do with unique single-family units, and
would have required a much bigger sample.
To get a sense of the rental market strength, I also looked at hotel rents. In
this case, most rooms are identical and the hotel charges the same rate for all
rooms for a given season.
Size of Sample
The sample consisted of five condominium projects with a total of 171
units, and four hotels with 239 rooms. All of the hotels were in existence in 1969
and the condominiums were built between 1967 and 1973.
12Based
on 1993 estimate by the Sherburne Town Clerk.
Geographic Regions Influencing Study Area
The assumption is that the primary geographic areas influencing the
Sherburne ski housing market were/are Connecticut, Massachusetts, New York
and New Jersey.
Economic Factors Affecting Study Area
The model assumes that the economic factors of primary importance are
the state employment figures, mortgage rates and the consumer price index.
Representative Characteristics of Study Area
Finally, the model assumes that Killington and Pico are representative of a
typical New England "Type II" destination ski resort, and the influencing factors
and trends should be similar.
Data Included
A summary of the basic data excluding sales price and rental data is shown in
Appendix Table A3.1
Time Line
All sales, stock and economic data are entered on a calendar year basis.
Rental data, skier visits and snowfall are seasonal data and are entered for the
spring year of the season (when most would have occurred). By doing this it is
possible to keep all the data in the same time frame for the sake of comparison.
The discrepancy between annual and seasonal data is eliminated with the inclusion
of time lags in the regression.
Condominium Sales
As mentioned previously, condominium sales were evaluated by studying
the sales and resales of 171 condominium units in 5 different condominium
projects. There were a total of 372 condominium sales over a period of 24 years
(see Appendix Table A3.2 for summary of sales data). The sales data was
collected from the Vermont transfer tax receipts kept on file in the Town Clerk's
office in Sherburne. The sales prices were only included for arms length
transactions in which a reasonable consideration was paid for transfer of
ownership. The sales prices are based on the value assigned to the basic real estate
excluding the assigned value of personal property sold with the units. The sales
prices are measured in nominal U.S. dollars.
Condominium Rents
Condominium rents for the 5 condominium projects tracked above are
based on winter season weekend rates. These rents were obtained from past
lodging directories kept on file by Killington Ski Resort (see Appendix Table A3.3
for a summary of condominium rental data). The lodging directories list rates for
the projects as they were brought on line for rental, starting in the early seventies.
There is some difficulty in tracking lodging rates over the years because the rate
structure changed at various times. Some years flat rates were charged for condo
units, while other years rates were a function of the occupancy. When rates were a
function of occupancy, the different units in some projects had to be figured
separately because of their different sizes and potential number of occupants. An
effort was made to try to compare the average annual rates for each of the units
over the course of the study period. Due to a lack of sufficient data for the first
couple of years in the early seventies, condo rental rates for this time frame were
artificially set against the hotel rates for those years. It should be noted that this
data is simply a measure of what was charged and does not reflect the occupancy
rate of the units. There is no measure of occupancy rates. Rents are measured in
nominal U.S. dollars.
Hotel Rents
Hotel rates for the four hotels tracked are the daily winter season room
rates over the last 25 years. This data was also obtained from the Killington
Lodging Directories. The ranges listed are for varying room sizes and for the
purpose of this study, average room rates were used (see Appendix Table A3.4 for
a summary of hotel room rates). These rates were easier to track than the condo
rates because the rate structure has remained constant over the years. Rates are
charged per room based on double occupancy and, except for The Cortina Inn, are
based on a modified American meal plan. The Cortina Inn rates are based on a
European meal plan. As with condo rates, the hotel rates do not measure
occupancy, presumably however, the rates are a reflection of the occupancy from
one year to the next. Rates are measured in nominal U.S. dollars.
Stock of Condominiums
The annual stock of condominiums in the town of Sherburne, Vermont is
shown in Appendix Table A3. 1. This data is an estimate based on the years in
which condominiums were first put up for sale and a survey conducted by the
Town Clerk in 1990 (see Appendix Table A3.5 for survey data) counting the total
number of condominium projects and units in existence at that time. This data was
used to back calculate the stock and number of completions in each year.
Number of Skier Visits
The number of skier visits is the total seasonal number of skier visits to the
State of Vermont as reported by the Vermont Ski Areas Association (see Table
A3.1 for data). A skier visit is classified as a day, night or partial day visit of a
single skier. If a multiple day pass is sold, each day that person skis is considered a
skier visit. Obtaining data about skier visits at specific resorts proved to be
impossible as this is a closely guarded trade secret. The results for the entire state
are better that those for Killington and Pico, however, since they show the trend in
skier visits across a spectrum of resorts. If we looked only at Killington visits and
tried to compare them to sales and rental prices we may get interdependence of the
results. As it is, however, we are evaluating the trend and eliminating the
singularities associated with Killington.
Regional Employment
Total non-agricultural state employment data is included for Connecticut,
Massachusetts, New York and New Jersey since, as mentioned previously, it is
assumed that these are the primary markets for a Vermont ski home. Vermont,
New Hampshire and Maine employment figures are not included since it is
assumed people in these states would not be in the market for a ski home in
Vermont. This data is summarized in Appendix Table A3.1 and was obtained from
the Employment andEarnings: States andAreas record published by the Bureau
of Labor Statistics.
Total Annual Snowfall
Total seasonal snowfall is included from the National Weather Service
weather observation station at Chittendon, Vermont (see Appendix Table A3.1 for
data). This data was obtained from the Northeast Regional Climate Center located
at Cornell University in Ithaca, New York. Chittendon is the nearest observation
station to Killington, approximately 10 miles north of the ski area, and is deemed a
good estimate of the natural snowfall at Killington. This does not account for
man-made snow on the mountains and may not reflect the actual snowfall on the
mountain, but it tracks the relative amount of snowfall in the region. Snowfall is
measured in Imperial inches.
Mortgage Rates
Mortgage rates are the annual national average charged by major banks as
listed in The Economic Report of the President(see Appendix Table A3.1 for a
summary). The real mortgage rates listed are the nominal mortgage rates minus
inflation rates for each year.
Consumer Price Index
The CPI is the annual figure as listed in The Economic Report of the
President(see Appendix Table A3.1 for a summary). The CPI is used to calculate
the inflation in each year by measuring the percent increase from one year to the
next.
Analysis of Data
Annual Price Index
In order to track trends over time it was necessary to convert the
voluminous amounts of sales data into annual price indices that would measure the
relative sales price from one year to the next across the full spectrum of projects
and units. This was done by performing a linear regression analysis on the sales
data and developing a hedonic model.13 By performing this regression it was
possible to develop an equation that can be used to predict the sales price of any
one of the units in the five condo projects by simply inputting the units
characteristics and the price index for the year in which the sale price is wanted.
By developing this equation, it is possible to derive an annual price index from a
group of dissimilar projects. The data from the five different condo developments
and from the different unit layouts within each development can be combined to
create a relative price index.
The data included in the price index equation is as follows:
Ln(Sales Price of Unit)
Square Feet of Unit
# Bedrooms in Unit
# Bathrooms in Unit
# of Floors in Unit
Dummy variable for each year from 1970 to 1992
Dummy variable for four of the five condominium developments
The sales price is tabulated as a logarithmic function in order to plot the
non-linear nature of sales prices over time.
13For a discussion of the
use of hedonic models see the article by Norman G.Miller.
The value of the dummy variables is either 0 or 1 depending on whether the
variable is false or true. The dummy variable for the years is used to develop the
price index. The first year of the study, 1969, is not included as a dummy variable
since it is the default value. When the regression is run it yields a regression
coefficient for each dummy variable which is used to derive the price index for that
year relative to 1969, which is indexed as 1.
The dummy variable for the condominium developments takes into account
the unique features of each. As with the year dummy variable, the condo dummy
variable has a default value as well. The default project is Edgemont (EM). The
other projects are Colony Club (CC), Hemlock Ridge (HR), Pico Townhouses
(PT) and Whiffletree (WT). They are located in different areas with different
vistas, amenities, layouts and so on. In the case of this study, the purpose is to
develop an annual price index and we don't so much care about how people value
the different amenities associated with condominiums. As a result we simply use
one dummy variable to account for all the unique features of each development. If
one was interested in how buyers value the different amenities, a separate
regression should be run with dummy variables included for each amenity.
The features square footage, # of bedrooms, # of bathrooms and # floors
are included since the units within a given project may have different
configurations.
The complete regression output with statistics is included in Table A2. 1.
The equation is shown in tabular format with the intercept indicated and the
coefficients for each of the variables shown under the heading "X Coefficient".
Analysis of this equation indicates that people value property more if it has greater
square footage, more bedrooms and more bathrooms - not very surprising. What
is interesting, however, is that people prefer a greater number of floors for a given
square footage, and are willing to pay for it. If a developer has the choice of
Table A2.1
Regression Output
9.832846223
0.11908105
0.934026635
366
334
Constant
Std Err of Y Est
R Squared
No. of Observations
Degrees of Freedom
Sq. Feet
Bedrooms
Baths
# Floors
D70
D71
D72
D73
D74
D75
D76
D77
D78
D79
D80
D81
D82
D83
D84
D85
D86
D87
D88
D89
D90
D91
D92
DCC
DHR
DPT
DWT
Std Err of Coef
X Coefficient
7.51103E-05
0.000214634
0.104761488
0.027079425
0.042031104
0.01644268
0.14689685
0.029278125
-0.2188155
0.064398394
0.058722127
-0.155420669
0.062264545
-0.078408072
0.056863024
-0.015478657
0.065353028
0.019438303
0.067533781
0.088444965
0.063771561
0.111572987
0.058348352
0.134073826
0.058414644
0.290982557
0.057686441
0.403892754
0.065957439
0.543767277
0.068414854
0.70724903
0.08118415
0.600061095
0.063301179
0.772225373
0.062818993
0.803011309
0.062941367
0.886096128
0.061222995
0.85136696
0.061109747
0.983962197
0.070482078
0.978327435
0.065866933
0.899822561
0.081210558
0.76097836
0.077449166
0.598835145
0.073951316
0.695866943
0.044280578
-0.357570914
0.028872123
-0.173345947
0.038221332
0.02148505
0.031712238
0.072937252
t Statistic
2.85758741
3.868674748
0.391202676
5.017290227
-3.397840958
-2.64671389
-1.259273198
-0.272209525
0.297435387
1.309640353
1.749572777
2.297816838
4.981328963
7.001519759
8.244214526
10.33765318
7.391357701
12.19922579
12.78293819
14.07812013
13.90599983
16.10155893
13.88051349
13.66121849
9.370436243
7.731976696
9.409797915
-8.075118477
-6.003921152
0.562121973
2.299971768
building horizontally or vertically, he/she will get greater value by building
vertically for the same square footage and number of units within a building
envelope.
Once we have this equation in terms of Ln(Sales Price), we can raise the
equation to the power of the natural log, e, and determine the annual price indices.
This data is shown in Table A2.2. The price index for 1969 then becomes 1 since
e0 is 1. All the years are then indexed to 1969 in nominal terms.
For the purpose of this study it is important to convert all these nominal
price indices into real price indices that can be equally compared with one another.
In order to do this, all the price indices were converted into 1992 dollars by
multiplying each annual price index by the 1992 CPI and dividing by the CPI for
the given year. With this real price index, we are now able to analyze the trends in
price over time and try to index it to other exogenous variables. A plot of this real
price index over time is shown in Figure Al .1. A look at this plot shows a peak in
the early seventies (when the economy was booming) followed by a steady decline
through 1977 (when the economy was experiencing a recession). The prices then
began to rise again reaching a peak in 1981 as the economy peaked out. The
region experienced a mild recession in 1982 with a reduction in jobs. Combined
with a huge increase in condo completions, this resulted in a significant drop in
prices in 1982. Real prices continued to rise from 1982 to 1987 as the economy
expanded with only a small drop in 1986. After 1987, however, real prices
dropped 40% over a period of four years as the economy slipped into a long
recession and the effect of the overbuilding of the 1980's sank in. Since the low in
1991, prices have risen a little, but they are still substantially lower than they were
six years ago.
Table A2.2
Price Index Calculation
Intercept
Sq. Feet
Bedrooms
Baths
# Floors
D70
D71
D72
D73
D74
D75
D76
D77
D78
D79
D80
D81
D82
D83
D84
D85
D86
D87
D88
D89
D90
D91
D92
DCC
DHR
DPT
DWT
9.832846
0.000215
0.104761
0.016443
0.146897
-0.21882
-0.15542
-0.07841
-0.01548
0.019438
0.088445
0.111573
0.134074
0.290983
0.403893
0.543767
0.707249
Exp. D
0.80347
0.856055
0.924587
0.984641
1.019628
1.092474
1.118035
1.143477
1.337741
0.600061
1.497643
1.722484
2.028404
1.82223
0.772225
2.164578
0.803011
2.232253
2.425642
2.342847
2.675034
2.660003
2.459167
0.886096
0.851367
0.983962
0.978327
0.899823
0.760978
0.598835
0.695867
-0.35757
-0.17335
0.021485
0.072937
2.140369
1.819998
2.005447
'Ut
0
1992
1990
1991
1989
1988
1987
(D
z
x
D
C).
o
OOI
---
-
-
1985
1986
-.
- -
-
PO1
1983
1984
1982
1981
1980
1976
1977
1978
1979
1974
1975
1973
1972
1971
1970
o
C
(D0
0
e/
ee ++
/\
l
\eI
K
\
0
Indices ($)
Annual Rent Indices
As with sales prices, it was necessary to convert the rental data from the
five projects and four hotels into annual indices for the purpose of analyzing
trends. In this case, however, a different approach was taken. Since data was
available for each project on a whole for each year (rates were established for all
condominiums in a project), a weighted average of the rents was used to determine
the relative rent movement from one year to the next. This weighted average was
calculated by summing the product of the total number of units in each project
times the rental rate and dividing the sum by the total number of units. A similar
calculation was made for hotel rents over the time period (see Tables A2.3a&b for
a summary of the calculations). For the condominium rents, the first four years
data was adjusted to more accurately reflect the changes measured in the hotel
market since the condo data for these years was incomplete.
As with the price indices, these indices were then converted from nominal
into real 1992 dollars for the sake of comparison. The same conversion process
was used as that described for prices using the CPI. A plot of the real rent index
over time is shown in Figure Al. 1 on page 34. This plot shows that the condo and
hotel rents follow similar trends over time, with some exceptions. Real rents
generally fell from 1970 through the mid-seventies during the energy crunch and
the recession of the mid-seventies. Rents then began a steady increase through the
late eighties with hotel rents spiking up to a record high from 1985 to 1988. Since
then rents have fallen off precipitously to around 70% of their late-eighties levels.
Table A2.3a
IYear
Year
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
Condominium Rents
Condo
Condominium Project
Rent
Whiffletree
Pico Townhouse
|Hemlock Ridge
Edgemont
|IEdaemont
IColony Club
Club
[Colonv
I Rent/we |I # Units I Rent/we I # Units I Rent/we | Index
| # Units | Rent/we | # Units | Rent/we| ## Units
Units I ent/we #Units I Rent/we
I # nits I Rent/we I
175
175
200
200
238
288
313
325
350
388
475
413
475
540
468
600
672
750
670
80
115
123
133
143
143
158
168
180
237
313
388
346
400
400
406
368
368
526
320
284
175
175
175
175
200
238
263
288
335
380
395
435
460
480
580
660
622
622
600
160
188
213
263
300
318
455
475
475
140
150
160
163
180
193
210
237
310
380
348
490
400
406
430
490
420
410
316
80
115
149.433
154.5294
166.1176
169.9381
192.1856
218.7216
239.5567
268.7526
322.2353
392.3299
391.3711
449.2887
423.4706
443.2471
444.1882
507.2941
523.8588
486.7529
419.3647
Adjusted
Value
130
124
128
133
Table A2.3b
Year
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
Hotel Rents
Hotel
Cortina Inn
linn @Long Trail
|Chalet Killington|
|Red Rob Inn
# Rooms IRate/day| # Rooms IRate/dayl # Rooms IRate/dayl # Rooms |Rate/day
{
36
39
39
42
46
50
49
52
60
70
80
92
96
100
120
125
125
123
150
160
160
141
130
38
41
42
45
58
60
56
64
64
70
75
84
97
110
90
75
86
98
116
120
120
120
120
90
90
90
90
90
90
90
90
90
90
90
90
90
90
90
90
90
90
90
90
90
90
30
32
30
32
33
35
41
45
54
60
65
70
75
88
89
115
152
180
140
151
148
150
40
39
40
40
39
43
47
48
56
67
69
88
98
108
126
134
144
153
158
158
158
168
168
Hotel
Rent
Index
37.68056
36.02564
37.17521
37.84188
42.5812
44.94017
45.04701
50.1453
54.65812
63.4359
69.58547
79.11966
86.51709
93.91453
101.2051
100.2051
114.2521
132.1667
154.3504
142.2906
146.5214
142.2692
140.453
------------
Skier Visit Regression Analysis
An analysis of the ski resort real estate market has to start with an analysis of the
skier market. It is intuitive that the more people there are skiing, the more people will
want to rent condos or hotel rooms, or buy real estate. The more people there are renting,
the higher the rents will be. The more people there are buying, the higher prices will be.
Therefore, one of the factors driving rents and prices is skier visits.
Numerous regressions were run for the skier visits equation, considering factors
such as the total employment in the four study states, the annual change in total
employment, the employment figures for each state separately and the seasonal snowfall
figures for the area. The best regression fit was found to include a lagged combination of
total regional employment and snowfall, with skier visits a function of the employment in
the current year, employment in the prior year, employment two years prior and the
snowfall in the current year.
The complete regression statistics for this equation are shown in Table A2.4. The
equation is shown in tabular format with the intercept and variable coefficients listed. The
R square of 0.893 indicates a very good correlation between the equation and the actual
results, which can be seen graphically in Figure A1.2. The favorable t-statistics from the
regression output also indicate that the results are statistically significant. The signs are
also as one would expect. The employment is predominantly positive, with a negative
correction, and the snowfall is positive. As employment increases or snowfall increases,
skier visits increase. This means that given employment projections and snowfall
forecasts, the model can accurately predict the number of future skier visits.
Table A2.4
Skier Visits
Regression Statistics
Multiple R
RSquare
Adjusted RSquare
Standard Error
Observations
0.945055084
0.893129112
0.867983021
354242.0836
22
Analysis of Variance
Regression
Residual
Total
Sum of Squares
1.78281E+13
2.13329E+12
1.99613E+13
Coefficients
Standard Error
t Statistic
P-value
Lower 95%
Upper 95%
-11813750.1
0.858473842
-0.579128925
0.657589231
19429.16566
1496499.058
0.308848207
0.557899766
0.323012655
4447.345474
-7.894258297
2.779597948
-1.038051923
2.035800212
4.368710678
1.01867E-07
0.011231301
0.311051359
0.054581502
0.000269055
-14971091.54
0.206860174
-1.756196185
-0.023908852
10046.07379
-8656408.67
1.510087511
0.597938336
1.339087314
28812.25753
Co fficie ts
Total Emp
Total Emp
Total Emp
Snowfall
Intercept
x1
x2
x3
F Significance F I
Mean Square
4.77946E-08
4.45701E+12 35.51761192
1.25487E+11
df
4
17
21
StandardError
fD
1992
1991
1990
1989
1988
1987
1986
1985
1984
1983
' 1982
1981
1980
1979
1978
1977
1976
1975
1974
1973
1972
1971
01
Skier Visits
Of course, however, the skier forecasts are only as good as the forecasted data put
into the equation. Employment data can probably be forecast with some accuracy by
economic experts given the economic indicators in a region. This data will be used in the
later chapter on forecasting. The real problem with this equation, however, is the
importance of snowfall and the lack of a reliable means of predicting seasonal snowfall in a
region. One could look at the Farmer'sAlmanac to get a sense of how much snowfall
might occur in a given year, but this is still very much a shot in the dark. It is very difficult
to predict with any accuracy what the snowfall several years in the future will be since it is
very much a function of climatological weather patterns which are constantly changing.
It is significant to note that skier visits are very well correlated with snowfall. Of
all the factors considered, the t-statistic for snowfall is the highest, 4.37, however, the
significance of the variable has been diminishing over the last decade. A plot of the
variables over time shows this trend. Figure A1.3 tracks the skier visits, snowfall, and
employment figures over the last 23 years. This figure yields some interesting results. Up
until the early eighties, skier visits in a given year closely followed the amount of natural
snowfall in that year, with peaks in 1971, 1978 and 1982, and troughs in 1974 and 1980.
After 1982, however, the relationship of visits and snowfall begins to separate with
snowfall trending downward and visits trending upward. This phenomenon is probably
the result of the introduction of snowmaking equipment at most Eastern ski areas. Up
until the early 1980's most areas relied upon natural snow and hence rode the weather
roller coaster from one year to the next. With the advent of modern snowmaking, ski
areas have been able to smooth the skier visit cycle somewhat.
The other interesting result is that prior to the early 1980's there was a smaller
relationship between skier visits and employment, in fact there was a reverse lag during the
1970's where employment followed skier visits by about one year. Whether this is
coincidence or not is unknown. Then starting in the early 1980's, the employment seems
to become more of a determining factor in predicting skier visits. The employment in the
1
I
1990.
1989
1988
1987
I
T
1984
1985
1986
/
+
II
1982
1983
1980
* 1981
1979
1977
1978
1976
1975
1974
1973
1972
1970
1971
I
94
D0
D
I | T
Employment
I
Skier Visits
||
I
I
current year seems to be the primary factor with the employment one and two years back
contributing to the equation. This may be the result of the increasing cost of skiing. Over
the last decade the cost of skiing has skyrocketed as equipment and lift expenses have
continued to increase. As a result of the increased cost, skier visits have become more
dependent upon employment.
Another factor which may affect skier visits and has been neglected from the study
is the effect of advertising and how the skier market has responded to it. How many
people ski and how has the number changed over the years? This data was not readily
available for study, but it seems that the popularity of the sport has substantially increased
over the years and is to some extent a function of demographic patterns. The skier market
is predominantly younger, affluent people aged 18 to 35 with incomes between $25,000
and $50,000.14 Over the last five years, there has a been a slowdown in the growth of the
skier market. In explaining this, Logan and Day say:
"A number of factors are responsible for this slowdown: fewer and shorter
visits by habitual skiers; a decline in leisure time; more competing choices
for that leisure time; a slow growth in the population of new skiers; and the
aging of the population... .The slowdown of growth in skier days is largely
a matter of changing demographic patterns. The aging of the baby
boomers is taking its toll on the skier market.....Although participation
rates among the older population are projected to improve over the next
few years, this improvement will not offset the loss in the number of skiers
under 35."15
Figure A1.3 on page 42 shows this considerable decline in skier visits over the last
several years from a peak in 1987.
In either event, there appears to be a strong correlation between employment and
skier visits as well as a diminishing yet still important correlation between natural snowfall
and skier visits. As time progresses and demographics change, it will be important to
14Logan and Day, "Ski Resort Real Estate Development", p.37.
15Logan
and Day, "Ski Resort Real Estate Development", pp. 37-38.
update this model to account for the changing demographic patterns. For the time being,
however, the model provides a good estimate of skier visits and will be used to forecast
skier demand.
I
~I~u
~~
-
~
~5
----------
Condominium Rent Regression Analysis
A glance back at Figures 1 and 2 on pages 20 and 21 shows the assumed factors
contributing to the determination of rents. Condominium rent is a function of the skier
visits, the condo stock, and the rent in the prior year. This would imply that rent is
indirectly a function of employment and snowfall.
In fact after running numerous linear regressions, it was determined that the
established condo rent is best described by a lagged function of visits, condo stock,
employment and the prior year's rent. Specifically, the equation is a function of last year's
visits, last year's stock, last year's rent, and the total employment this year.
The complete regression statistics for this equation can be seen in Table A2.5. The
equation is shown in tabular format with the intercept and variable coefficients listed. The
R square of 0.84 indicates a good statistical fit which can be seen in Figure Al.4. The tstatistic values indicate that the variables are statistically significant for the 90%
confidence interval and all but the visits are statistically significant for the 95% confidence
interval. The signs are correct, indicating that when visits or employment increase the rent
will increase, and that when stock increases rents will decrease. The coefficient of
condominium rent indicates that 78% of last year's rent automatically contributes to this
year's rent. Analysis of Figure A1.4 indicates that while the value of rent was not
absolutely predicted in some years such as 1980, 1983, 1985, 1988 and 1989, the trends
of the rental curve are accurately predicted.
The real question in this case is whether owners actually consider all these factors
when deciding rents for a given year. In all likelihood the answer is no. They probably
look back at the past year's occupancy and try to decide whether to raise, lower or leave
Table A2.5
Condominium Rent
Regression Statistics
Multiple R
R Square
Adjusted R Square
Standard Error
Observations
0.914032771
0.835455906
0.796739649
33.85717265
22
Analysis of Variance
Regression
Residual
Total
F
df Sum of Squares Mean Square
4 98944.47137 24736.11784 21.57894286
17 19487.23837 1146.30814
21 118431.7097
Significance F
1.76587E-06
P-value
Lower 95%
Coefficients Standard Error
Visits -1
Stock -1
C Rent -1
Tot EmpI
Intercept
x1
x2
x3
x4
-432.166425
2.27595E-05
-0.09923506
0.782080483
3.55599E-05
t Statistic
Upper 95%
210.5623659 -2.05243906 0.052799097 -876.4148051 12.08195526
1.81813E-05 1.251806794 0.224404587 -1.55998E-05 6.11187E-05
-2.2272117 0.037003727 -0.193239588 -0.00523054
0.044555739
0.264076199 2.96157127 0.007446943 0.224927627 1.33923334
1.80213E-05 1.973221667 0.061775935 -2.46165E-06 7.35815E-05
1992U
1991
1990
1988 -
1987 --
1986 -
1985 --
1984 -
1983 -
1982
1981
1980
1979 -
1978
1977
1976
1975 -
1974
1973
1972
1971 -
oW
-
CD
a.
1 -(D
(D
7
Real Condo Rent
-a
3
0
0
C.
0
the rents unchanged. The occupancy would be their gauge of demand and would enable
them to determine if their rents were too high or too low.
The author proposes that this is backwards thinking and the model is taking this
into account when it tries to fit an equation to the data. The statistical equation forecasts
the net rent that equals the demand for a given stock based on historical data.
Conceptually, this year's rental data should have no relation to the past year's skier visits as
it seems to have. Rather, this year's rental data should be a function of the projected skier
visits for this year. The regression statistics for an equation assuming this year's skier
visits determine this year's rental rates is shown in Table A2.6. The t-statistic for skier
visits is negative and insignificant. This either confirms the suspicion that this years
projected skier visits are not used to determine rent or, if projections were used, indicates
that the projections were completely inaccurate. An analysis considering both this year's
and last year's skier visits yields similar results.
In fact it may simply be that the owners' representative sets the rents in any given
year based on the performance of the previous year. After all, it is in the interest of the
agent to see that all units get rented since that means increased commissions. The agent
knows what the occupancy was the year before and may use that at a gauge of whether
the rent was too high or too low. This results in the classical owner - agent dilemma.
The author proposes that last year's skier visits can be used in conjunction with last
year's occupancy rates to determine the relative pricing of a given property. This should,
however, be used with a prediction of this year's skier visits to set a rent level. For
example, if last year's skier visits were relatively low, occupancy was high, and skier visits
are projected to be significantly increased this year, a significant increase in the rent would
be appropriate. If, on the other hand, we based this years rent on the low skier visits last
year and the high occupancy we would be more likely to increase rents only slightly. This
lack of foresight can lead to inefficient pricing in the rental market.
Table A2.6
Condominium Rent
Regression Statistics
Multiple R
R Square
Adjusted R Square
Standard Error
Observations
0.914633843
0.836555067
0.785478525
34.78241552
22
Analysis of Variance
Regression
Residual
Total
Intercept
x1
Visits
Visits -1 x2
Stock -1 x3
C Rent -1 x4
Tot Empi x5
F
df Sum of Squares Mean Square
16.37845949
19814.92938
5 99074.64688
16 19357.06287 1209.816429
21 118431.7097
Significance F
8.51026E-06
Coefficients
Coeffidents
Standard Error
Error
t Statistic
P-value
Lower 95%
Upper 95%
-454.098666
-5.2065E-06
2.23851 E-05
-0.09705991
0.785422104
3.81736E-05
226.4141405
1.58724E-05
1.8713E-05
0.046251171
0.271484025
2.01556E-05
-2.00561089
-0.32802345
1.196233711
-2.09853958
2.893069325
1.893947797
0.057953267
0.746141974
0.244937488
0.04813012
0.008700163
0.072089873
-934.0750941
-3.88545E-05
-1.72847E-05
-0.195107995
0.209901811
-4.5543E-06
25.87776174
2.84415E-05
6.20549E-05
0.000988168
1.360942397
8.09015E-05
Standard
Ideally, the model should be able to predict ahead of time what the demand will be
in a given year irrespective of the demand last year, and set the rent level accordingly. It
should pre-determine this year's performance instead of only working backwards off last
year's performance. In this regard the current model is faulty and should be corrected over
time.
Hotel Rent Regression Analysis
Hotel rent was included as another means of gauging the demand for resort
accommodations. The same rate structure has been used over the years and hence it was
easier to track the variation in hotel rents than it was to track the condo rents. The
numerous linear regressions that were run found the hotel rent to be primarily a function
of skier visits and the hotel rent the prior year. Data about stock of hotel rooms was
unavailable and hence was not considered in the regression. Unlike condo rent, hotel rent
was found to be very well correlated with skier visits in the current year, indicating that
hotel operators have some basis for estimating the demand for the upcoming year.
The regression statistics for this equation are shown in Table A2.7. The equation
is shown in tabular format with the intercept and variable coefficients listed. The R square
of 0.88 indicates a very good correlation between the actual and projected data. This fit
can be seen graphically in Figure A1.5. All the variables are statistically significant at the
90% confidence interval and all except last year's skier visits are significant at the 95%
confidence interval. The significance of this equation is that it shows that hotel operators
make an estimation of skier visits in the approaching year and base rates more on this
projection than they do on the previous year's attendance. They are looking forward
rather than simply bench marking off the past. As a result they are operating more
efficiently than the condo market and should be experiencing greater revenues. A plot of
real hotel rents versus skier visits is shown in Figure A1.6. This shows that while there are
Table A2.7
Hofel Rent
Regression Statistics
0.93961613
Multiple R
RSquare
0.882878471
Adjusted RSquare 0.863358217
7.687111181
Standard Error
22
Observations
Analysis of Variance
df
Regression
Residual
Total
Intercept
xl
Visits
Visits -1 x2
H Rent -1 x3
rum of Squares Mean Square
F Significance F
3 8017.944111 2672.648037 45.22883955
18
1063.65021 59.09167831
21 9081.594321
1.38218E-08
P-value
Lower 95%
Coefficients Standard Error
t Statistic
12.29850796
2.80019E-06
3.38869E-06
0.153950688
2.660709835
2.623870066
1.485395692
2.821801429
32.72276108
7.34733E-06
5.03354E-06
0.434418271
Upper 95%
0.014627468 6.884534649 58.56098751
0.015863366 1.46435E-06 1.32303E-05
0.152297183 -2.08583E-06 1.21529E-05
0.010217223 0.110979628 0.757856915
eD1981
1992I
1991
1990 -
1989 --
1988 -
1987 --
1986 --
1985 -
1984 -
1983 --
1982 -
8~
1980
1979
1978
1977
1976
1975
1974
1973
1972
1971
I
0
(
-
(D
0.
19(D
Q
Real Hotel Rent Index
09
In
C-
c
0
Z66 L
L66L
066L
696L
996L
L96 l
996l
996 l
P96 L
£961L
Z96 L
1961
0961L
6L6 l
9L6 L
LL6l
9L6 l
t7L6 L
CL6 I
ZL6 I
LL6I
0L6I
I-
some lags between changes in hotel rents and skier visits, there is a general trend for hotel
rents to change with skier visits.
The explanation for the improved efficiency in the hotel versus condo markets is
that hotels are operating businesses that make their livelihood off of renting. They have to
be efficient if they want to remain in business. On the other hand, condo's are typically
owned as second-homes for private use and some people choose to reduce the costs by
renting out the units when they are not there. They are less concerned with getting into
the operating details of the business and are happy just to be getting an occasional rent
check in the mail.
Summary
The conclusion of the rental market aspect of the study is that reliable equations
can be and were developed to predict the rental rates in both the condo and hotel markets.
These equations are reliant upon the previous equation predicting skier visits, which in
turn is a function of the total regional employment. Rents are therefore a function of total
regional employment.
It was found that the condo rental market is operating inefficiently and is basing
rates on last year's attendance rather than predictions of this year's attendance. Hotels on
the other hand have been acting proactively by setting rates based on projected attendance
in the coming year. This may be the result of the owner - agent dilemma discussed in
elementary economics. In the case of the hotel, the owner is the agent and it is in his/her
best interest to maximize the profits by charging the highest rates the market will bear.
Conversely, in the condo market the owners are represented by brokers who get a
commission by renting units. Depending on the commission structure, the broker may
receive more compensation for renting at higher rates, but they may be unwilling to take
the chance of a reduced number of rentals. The broker would prefer to rent as many units
as possible to maximize their profits. If they charge less than what the market is willing to
pay they can be assured cash flow. A reverse psychology may develop where brokers
remember how they did the prior year and suggest this year's rates based on last year's
performance.
The key thing to remember about the rental market is that it is a function of the
demand, which in this case is skier visits. Skier visits in turn are a function of the
economy and the snowfall, although the economy is increasingly more important that
snowfall in New England. It is imperative that one know where the market is coming
from and analyze economic trends, specifically employment, in the regional market area in
order to predict variations in the rental market.
Condominium Price Regression Analysis
As with the rental market, the asset market is a function of the skier visits
to a resort. The more skier visits, the greater the demand will be and hence the
higher the prices will be. The asset market, however, is different from the rental
market in that people buy for different reasons than they rent, and different factors
influence the prices than influence the rents.
In analyzing the asset market, two separate models were compared for
buyers. It was assumed that buyers were either primarily interested in using the
second home as an investment tool, or were primarily interested in using the
second home for their own personal use. These two models have been classified as
the investment model and the consumer model.
In each case, the real price index is used in place of the actual
condominium price. To determine a specific condominium price the index has to
be converted back as discussed earlier.
Investment Market
The investment model is based on the assumption that someone will buy a
ski resort condo for the investment potential, namely the cash flow and the
expected capital appreciation. The traditional "Gordon Growth Model" states that
an asset's value is equal to the capitalized cash flow of the asset. The capitalization
rate should be the interest rate minus the growth rate. In this case the investment
equation in real terms would look as follows:
Real Price =
Real Rent
(Mortgage Rate - Inflation Rate) - A Real Rent
Numerous linear regressions were run with these variables using various
lags in an attempt to derive the classic investment model. The results were less
than extraordinary, indicating that prices in general are not established purely by
the traditional means of investment valuation.
The regression statistics for the best pure investment equation derived are
shown in Table A2.8. The equation is shown in tabular format with the intercept
and variable coefficients listed. The last term is this equation is the annual
mortgage rate minus the average inflation rate for the previous three years (a more
accurate representation of investors inflation expectations). The R square of 0.72
indicates a fairly good fit and the t-statistics indicate the variables are all significant
at the 90% confidence interval, with most being significant at the 95% confidence
interval. The plot of the curve also shows a good fit (see Figure Al.7). The
problem is that the equation does not make sense based on the signs of the
coefficients. The coefficient for last year's rent is negative indicating that if the
rent goes up the price should go down. The other problem is the mortgage rate.
The coefficient is positive, indicating that if the real mortgage rate goes up, the
price should go up. An increasing real mortgage rate indicates a decreasing
inflation rate and less likelihood that people would invest in real estate. The
coefficient for mortgage rates should be negative. This leads one to conclude that
the ski condo market can not be characterized by the traditional investment model.
It should be noted that the lack of a statistical fit for the investment model is not a
lag problem. All the variables were studied with various lag combinations and the
results were still inconclusive. The statistics for all these other equations have
been omitted due their volume and meaningless results.
One might suspect the reason for the difficulty with the investment model is
the inclusion of data after 1986, when the Tax Reform Act changed the tax status
of investment real estate. The fact, however, is that an analysis including only data
Table A2.8
Real Price Index
Regression Statistics
Multiple R
R Square
Adjusted R Square
Standard Error
Observations
0.849742156
0.722061732
0.647944861
0.226863549
20
Analysis of Varance
df
Regression
Residual
Total
4
2.005610924
15
19
0.77200605
Real Price Ind -1
Real C Rent Ind -1
d Real C Rent Ind -2
Avg. Real Mortg. Rate -1
x1
Standard Error
1.991223242
0.728511141
0.81504765
0.183008072
0.00151943
0.817546876
6.510328673
x2
x3
x4
-0.003812665
1.088513816
12.6054364
Mean Square
F
0.501402731
0.05146707
9.742204693
Significance F
0.000433352
2.777616974
Coefficients
Coefficients
Intercept
Sum of Squares
StandardError
t Stahstic
2.443075864
3.980759601
-2.509272598
1.331439026
1.9362212
P-value
0.024501347
0.000800618
0.021319327
0.198799587
0.067859771
Lower 95%
Upper 95%
0.253989231
3.728457253
0.338438429
1.118583853
-0.000574074
2.831074805
26.48188202
-0.007051256
-0.654047173
-1.271009227
1992
1991
1990 -L-(D
1989 - -CL
1988 -
1987
1986
1985
1984
1983
1982
1981
1980
1979
1978
1977
1976
1975
1974
1973
=
a
O 0 CL
o
I
Real Price Index
C,,
C)
prior to 1986 has an even worse statistical fit (see Table A2.9). This leads us to
the conclusion that the price of ski resort real estate can not be estimated by
considering only rent, mortgage rates and inflation.
The next extension to add to the investment market model is the inclusion
of employment. Presumably people would not be making investments in second
homes unless they had primary jobs to support themselves. As it turns out,
including employment figures does improve the model, however, there are still
significant problems. Adding employment to the previous model and running
numerous lagged regressions led to only one combination of variables that resulted
in the correct coefficient signs. This does not inspire much confidence in the
results since it might be a fluke rather than a real trend.
The complete regression statistics for this equation are shown in Table
A2. 10. The equation is shown in tabular format with the intercept and variable
coefficients listed. The equation is shown graphically in Figure A1.8. While the R
square has improved to 0.75 and the signs are correct, the t-statistics for the rent,
change in rent and mortgage rate have dropped considerably to the point where
they are no longer statistically significant. As with the prior investment equation,
this equation is suspect and should not be used. The interesting lesson to be
learned from this is that the ski condo market can not be defined as an investment
market, contrary to what might be expected. The market defies the traditional
investment valuation rules. This leads one to conclude that all the investors are
acting irrationally or there are other factors involved in the price determination
(most likely the latter).
Consumer Market
The alternative model is the consumer model. This assumes that people
buy ski resort real estate because they like to use it. This model is in fact a better
representation of the pricing mechanism as will be shown. To predict price
Table A2.9
Real Price Index - Analysis Prior to 1987
Regression Stafistics
Multiple R
R Square
Adjusted R Square
Standard Error
Observations
0.795088699
0.63216604
0.46868428
0.143108669
14
Analysis of Variance
Regression
Residual
Total
Intercept
x1
Real Price Ind -1
x2
Real C Rent Ind -1
x3
d Real C Rent Ind -2
Avg. Real Mortg. Rate -1 x4
Z
0. 04268526
P-value
Lower 95%
4.019570652
0.001457398
0.279071797
0.001297578
0.652163932
-0.434803978
-0.737778344
1.202582491
0.670835327
0.473757018
1.450955348
-0.752646274
-0.003892654
0.25058438
-0.691017509
2.25957936
4.536244934
2.100207841
0.055791553
-0.73464961
19.78876397
Sum of Squares
Mean Square
0.316777068
0.18432082
0.501097888
0.079194267
0.020480091
Coefficients
Standard Error
Coefficients
3.318640317
-0.121341528
-0.000957325
0.784280926
9.52705718
Significance F
F
3.86689034 9
df
4
9
13
StandardError
0.825620596
t Statstic
Upper 95%
5.186325285
0.509963219
0.001978003
Table A2.10
Real Price Index
Regression Statistics
Multiple R
R Square
Adjusted R Square
Standard Error
Observations
0.867449968
0.752469447
0.669959263
0.214847567
21
Analysis of Variance
Regression
Residual
Total
Real Price Ind -1
Real C Rent Ind -2
d Real C Rent Ind
Avg. Real Mortg Rates
d Total Employment
Intercept
x1
x2
x3
x4
x5
df
Sum of Squares
5
15
20
2.104806608
0.692392155
Coefficients
Standard Error
1.056060241
0.548827948
0.000492766
0.773975019
0.688642093
0.194442186
Mean Square
0.420961322
F
Significance F
9.1197160 1
0.000380886
t Staffsfic
P-value
Lower 95%
1.533540066
0.140810249
2.822576509
0.420801535
0.963451286
-0.522256446
2.952903476
0.010518017
0.678387268
-0.411746538
0.134383983
-0.002003202
0.046159477
2.797198764
0.001171018
-2.546462607
0.803335914
4.875885452
8.412674838
2.848950162
0.346826515
0.607228654
0.007865666
-0.938296003
-12.93917283
2.340277576
Upper 95%
2.52386702
0.963271912
0.002988734
2.486246041
7.846247614
14.4850721
1992
1991
1990
1989 -5-
1988 -
1987
1986
1985
1984
1983
1982
1981
1980
1979
1978
1977
1976
1975 -
1974
1973
1972
I
o
0CT'
X
CD
(D
-1--
C
x
(D
a
A.
0x
CL
a
-o
-
~PO
I
I
I
Real Price Index
E9
l
Co~CT
movements in the consumer market two similar models have been developed.
Both models rely on historical price, condominium stock and change in regional
employment. In addition, the first model considers skier visits while the second
considers total employment. Conceptually this is a minor difference since visits are
a function of total employment, however, the reason for the two models will
become apparent.
The first model states that real condominium prices are a function of last
year's price, the skier visits, the condominium stock and the change in employment.
The complete regression statistics for the best fitting model are shown in
Table A2. 11. The equation is shown in tabular format with the intercept and
variable coefficients listed. The plot of the equation is shown graphically in Figure
A1.9. The R square of 0.72 indicates a fairly good statistical fit between the actual
and predicted results. The t-statistics for the variables indicate they are all
significant at the 90% confidence interval and they all have the expected sign.
Except for skier visits, which is nearly 95% significant, the other variables are all
significant at the 95% confidence interval. In the above equation, skier visits are
input for the next year due to the data format. Sales are catalogued according to
the calendar year and skier visits are recorded according to the seasonal year and
input in the year corresponding to the spring season. From the data, most people
bought houses towards the end of the calendar year, primarily in the last three
months. After buying a house, these people proceeded to ski and their visits were
recorded in the following year. This leads us to conclude that increased skier visits
are associated with increased buying activity and increased prices, as expected.
The condominium stock two years ago was found to have the greatest
impact on current prices of all the stock lags considered. This may be due in part
to the recording mechanism of the stock. Each condominium complex was
assumed to have its full number of units built and available in the year in which the
Table A2.11
Real Price Index
Regression Statistics
Multiple R
R Square
Adjusted R Square
Standard Error
Observations
0.849594813
0.721811347
0.652264183
0.191142836
21
Analysis of Variance
df
Regression
Residual
Total
Real Price Ind -1
Skier Visits + 1
Condo Stock -2
d Total Emp
Intercept
x1
x2
x3
x4
Sum of Squares
4
16
20
1.516772085
0.584569341
2.101341426
Coefficients
Standard Error
0.530594947
0.700319507
1.23777E-07
-0.000334618
5.312476437
0.596825792
0.204024874
7.37958E-08
0.000140501
2.333778692
Mean Square
0.379193021
0.036535584
t Stausuc
0.889028179
3.432520226
1.677285687
-2.381605464
2.276341135
F
10.37873168
P-value
0.384556396
0.002635933
0.109047244
0.027283809
0.033969938
Signincance F I
0.000242992
Lower 95%
Upper 95%
-0.734618926
0.267806193
-3.26635E-08
-0.000632468
0.365087737
1.79580882
1.132832821
2.80217E-07
-3.67693E-05
10.25986514
1991
1990
1989
1988 -L
1987 -.
1986 - -
1985 -
1984 -
1983 -
1982 -
~1981K
1980
121
1979
1978
1977 -
1976
1975 -
1974
1973
1972 -
1971
I
o
0
CD
D~
a
I
-
I
I
Real Price Index
e
0
3
0
0
condominium was legally declared. In some of the projects, the full number of
units may not have been immediately available. The bulk of the sales in the
individual complexes tended to lag behind the condo declarations. Another
possible reason for the lag would be if a purchase and sale agreement was reached
in the calendar year previous to the closing, when the recorded stock was lower.
Finally, the price market may simply be slow to respond to changes in the
stock. A real estate market is clearly not as efficient as a stock market and it takes
a longer time for the market to reach equilibrium. Surges in supply do not result in
instantaneous drops in values. In the case of the Killington real estate market, it
takes two years for the market to reach equilibrium from changes in the supply of
units.
As the model shows, the price market (as with the skier and rental markets)
is very sensitive to changes in employment. When employment in the buyer region
declines, the price of housing declines due to a lack of people buying second
homes. When financial security at home is threatened, people naturally retreat
from vacation and leisure expenses (as could be expected).
Finally, the consumer model is dependent upon the sales price in the prior
year as an index of where to price from. In this model, the index starts with
roughly 70% of the prior years value. The effects of the changed conditions over
the past year are then added and subtracted from this value to establish the current
year's price index.
The second consumer model is similar to the first except that it replaces
skier visits with total employment in the current year. The regression statistics for
this equation are shown in Table A2.12 and the plot is shown graphically in Figure
Al.10. The equation is shown in tabular format with the intercept and variable
coefficients listed. The advantage of this equation is a slightly improved R square
Table A2.12
Real Price index
Regression Statistics
Multiple R
R Square
Adjusted R Square
Standard Error
Observations
0.875864423
0.767138487
0.712347542
0.195742919
22
Analysis of Variance
df
Regression
Residual
Total
Real Price Ind -1
Condo Stock -2
Total Employment
d Total Emp
Intercept
x1
x2
x3
x4
Sum of Squares
4
17
21
2.145838828
0.651359937
2.797198765
Coefficients
Standard Error
-0.577818228
0.451084322
-0.000480834
1.61026E-07
4.933556584
1.423174315
0.195372732
0.00029728
1.27312E-07
2.647442967
Mean Square
0.536459707
0.03831529
t Statistic
-0.406006644
2.308839716
-1.617445913
1.264821705
1.863517608
F
Significance F
14.0011911 9
3.13753E-05
P-value
Lower 95%
Upper 95%
0.688845079
0.031218516
0.120705429
0.219791896
0.076434208
-3.580457762
0.038883314
-0.00110804
-1.07578E-07
-0.652067629
2.424821306
0.863285331
0.000146372
4.29631 E-07
10.5191808
III
oo5
-
1992
(D
1991
X
5D-
0
(
CL)
-~(D
09
o
1990
1989
1988
1987
1986
1985
1984
1983
1982
1981
1980
1979
1978
1977
1976
1975
1974
1973
1972
1971
91
C-
A3
Real Price Index
CCA0
of 0.77, however, the t-statistic for total employment is lower and just misses
being statistically significant at the 90% confidence interval. The primary
difference between this equation and the prior one is that this equation makes a
direct connection between employment and prices, skipping the intermediate skier
visit factor. This assumes that not all buyers of ski resort real estate are skiers.
People may simply like to get away to the mountains, maybe one member of the
family skis, perhaps people see it as an opportunity to bring extended family
together, or perhaps people see it as an investment for the future. While we
discovered in the investment section that people do not value condos with the
traditional cash flow model, they may see the real estate as an investment in a
retirement home or may simply be expecting appreciation in value, using the condo
as a long term hedge against inflation.
Which model is better? It should be obvious at this point that the
consumer model seems to better define the ski resort real estate market than the
investment model. Prices are more a function of the prior year's price index, the
condo stock, the regional employment and skier visits than they are a function of
rent, mortgage rates or inflation. In general people do not consider second homes
to be investment tools in the traditional sense. Of the two consumer models
considered they both seem to provide relatively good statistical fits with the data.
The data for the skier model is more statistically significant, yet the employment
model has a slightly better correlation. An advantage of the employment model is
one less variable to estimate by use of another regression. Rather than using the
skier visit equation and including the result, employment can be input directly. In
practice, both models could be calculated separately and an average of the two
used.
Condominium Construction Regression Analysis
The last model to be developed is that for condominium completions. The price
models developed above were dependent upon the stock of units in a given year. In order
to predict the stock, it is necessary to estimate the number of completions in a given year.
As will be shown, this can be rather difficult. It was difficult to develop a good predictive
model since some years in the data set contained zero completions while other years
contained several hundred completions, due to the nature of construction. Completions
were found to be a function of rent and price, with price being more significant. After
running numerous linear regressions, the best equation was related to last year's real condo
rent, last year's change in the real condo rent and last year's change in the real condo price.
The complete regression statistics for this equation are shown in Table A2.13 and
the plot is shown graphically in Figure Al. 11. The equation is shown in tabular format
with the intercept and variable coefficients listed. The R square of 0.42 signifies the
relatively low correlation of actual with predicted values, again largely due to the
excessive variation in the number of completions from one year to the next. The figure
does, however, show that the model predicts the general trends well although the curve is
somewhat smoothed. The t-statistics indicate that last year's change in price is the most
significant factor while the rent terms are less significant. This makes sense given the
information discussed thus far. As developers sense an increase in prices in the real estate
market they are more likely to start building. If last year's prices showed an increase over
the prior year, this would have shown increasing demand in the condo market. If the rent
also increased last year over the prior year this would have shown an increase in the
number of skiers and hence increased demand. Developers would have sensed the
opportunity and would have started to build, leading to completions in this year.
A plot of completions, rent and price is shown in Figure Al.12. This plot is
instructive in showing the movement of price with completions and rent. In this chart,
Table A2.13
Completions
Regression Statistics
Multiple R
RSquare
Adjusted RSquare
Standard Error
Observations
0.644871527
0.415859286
0.31277563
92.3579408
21
Analysis of Variance
Regression
Residual
Total
Intercept
R Condo Rent -I x1
x2
d R Pr. Ind. -1
d RCondo Rt -1 x3
df
3
17
20
F Significance F
Sum of Squares Mean Square
U.U24532362
34411.6166 4.0341 922e 9
103234.8498
8529.989228
145009.8169
248244.6667
Coefficients
('-o ffici ts
Error
Standard
StandardError
t Statistic
P-value
-92.42653191
0.348664069
215.845996
0.812614263
137.3397557
0.290866436
88.18848759
0.635745412
-0.672977256
1.198708501
2.447552985
1.278207044
0.508663924
0.244652394
0.023740126
0.215812308
Lower 95%
Upper 95%
-382.1884926 197.3354288
-0.265011325 0.962339463
29.7842913 401.9077008
-0.528693183 2.15392171
po
1992
1991
1990
1989
1988
1987
1986
1985
1984
1983
8 1982
1981
1980
1979
1978
1977
1976
1975
1974
1973
1972
3
>
Annual Condo Completions
0
o
3n-9
0
D
D.
C0.
Price Index, Rent Indices &Completions
250
Completions/2
-
200
-
Real Price Index*30
-
Real Condo Rent Index/4
-
Real Hotel Rent Index
150
100
50
0
0-
-
-
-
CV)-
. --
LO
- ---
- .0
r-
C-
C--
Year
Figure A1.12
'0
demand is represented by rent and supply is represented by completions. The greater the
rent, the more people there are visiting the area. The more people there are visiting, the
greater the chances people will want to buy. As the chart shows, as demand increases and
stock stays the same, prices increase. When demand decreases and stock increases, prices
decrease. When demand and supply move together a balance is reached and price either
goes up or down. An example of this in shown in 1982. The rent increased yet the price
decreased due to a surge in completions.
Summary
Of the two models considered (investment and consumer) the consumer model
seems to provide the best statistical prediction of prices. The actual model, however, may
be an intangible combination of both. People buy second homes when there is growing
employment in their primary region and when visits to ski resorts are increasing. They pay
a price which is a function of the number of buyers and the number of potential sellers.
They buy the real estate without concerning themselves with the mortgage rates or
inflation and not much consideration of the rental rates. Some people choose to rent their
units while others choose not to. While most people invest in second homes for their
personal use, they are still expecting the unit to appreciate in value over the years. In this
sense, the market could be characterized as an investment market, although the buyers are
primarily consumers. These buyers are not sophisticated investors on the whole and are
not valuing the property based on the projected cash flow.
M
-----
-----
With all these models in hand, it is now possible to run forecasts for the future of
the skier market, the rental market and the asset market. These forecasts will be based
primarily on projections for the growth in employment. Snowfall projections are included
for the visits equation. Consumer price index assumptions could be included to convert
from real to nominal dollars.
Employment projections are based on forecasts by Terleckyj and Coleman.16
These projections are made for the years 1995, 2000 and 2010. For the purpose of
modeling, a linear interpolation was made between these years. These projections were
used to calculate expected growth rates for each state which were then applied to the
existing state employment data. This data is shown in Table A2.14. Optimistic and
pessimistic variations on these projections are included (assuming annual growth rates
0.75% higher and 0.75% lower than those projected). It should be noted, however, that
the economists never forecast drops in employment and did not predict the drops in
employment that occurred in the 1990's. Even the pessimistic employment projections
assume a small positive growth rate over time. For summary plots of the employment
projections for each state and the regional total see Figures Al. 13a-e.
The real key to using these models is to be able to predict with accuracy when the
employment will rise and when it will fall from one year to the next. If this can be done
accurately, the models will be very profitable to the investor.
Snowfall projections assume that seasonal snowfall will return to the long term
average of the past twenty years over the next decade and a half This would be an
16Nestor
E. Terleckyj and Charles D. Coleman, Regional Economic Growth in the United States:
Projectionsfor1991-2010 (Washington D.C.:NPA Data Services, 1991), Vol. 1, pp. 49-51.
Table A2.14
Economic Projections for 1991 - 2010
199Q2
State
State
CT
MA
NY
NJ
Ann. Growth
Tot. Empl.
Ann. Growth
Tot. Empl.
Ann. Growth
Tot. Empl.
Ann. Growth
Tot. Empl.
1,506,500
2,750,500
7,650,200
3,386,800
1993
1993
1.61%
1,530,721
1.52%
2,792,191
0.93%
7,721,682
1.67%
3,443,467
1994
1994
1.61%
1,555,332
1.52%
2,834,513
0.93%
7,793,831
1.67%
3,501,082
1998
1.37%
1,646,101
1.37%
2,996,942
0.90%
8,081,361
1.42%
3,713,836
1999
1.37%
1,668,624
1.37%
3,037,856
0.90%
8,154,224
1.42%
3,766,697
2000
1.37%
1,691 A56
1.37%
3,079,328
0.90%
8,227,744
1.42%
3,820,311
1997
1998
1996
1996
2.12%
2.12%
2.12%
1,720,451
1,684,763
1,615,592 1,649,815
2.12%
2.12%
2.27%
2.12%
2,941,726 3,003,949 3,067,488 3,132,371
1.65%
1.65%
1.65%
1.68%
8,043,322 8,176,167 8,311,206 8,448,476
2.17%
2.17%
2.17%
2.42%
3,881,479
3,798,915
3,639,018 3,718,107
1999
2.12%
1,756,896
2.12%
3,198,626
1.65%
8,588,013
2.17%
3,965,838
2000
2.12%
1,794,112
2.12%
3,266,283
1.65%
8,729,854
2.17%
4,052,030
1999
0.62%
1,584,182
0.62%
2,884,061
0.15%
7,739,353
0.67%
3,576,189
2000
0.62%
1,593,977
0.62%
2,901,804
0.15%
7,751,088
0.67%
3,600,270
1995
1995
1.61%
1,580,338
1.52%
2,877,A77
0.93%
7,866,655
1.67%
3,559,662
1996
1.37%
1,601,962
1.37%
2,916,760
0.90%
7,937,582
1.42%
3,610,328
1997
1.37%
1,623,881
1.37%
2,956,579
0.90%
8,009,149
1.42%
3,661,716
Optimistic Projections --- > 0.75% higher than expected
1994
1995
Staite
1993
199
1995
1994
State
CT
MA
Ann. Growth
Tot. Empl. 1,506,500
Ann. Growth
Tot. EmpI. 2,750,500
NY
Ann. Growth
NJ
Tot. Empl. 7,650,200
Ann. Growth
Tot. EmpI. 3,386,800
2.36%
1,542,020
2.27%
2,812,819
1.68%
7,779,058
2.42%
3,468,868
2.36%
1,578,377
2.27%
2,876,551
1.68%
7,910,087
2.42%
3,552,925
2.36%
Pessimistic Projections --- > 0.75% lower than expected
1992
1993
1994
1995
State
1995
1994
1993
1992
State
CT
MA
NY
NJ
Ann. Growth
Tot. Empl.
Ann. Growth
Tot. Empl.
Ann. Growth
Tot. EmpI.
Ann. Growth
Tot. EmpI.
0.86%
1,506,500 1,519,422
0.77%
2,750,500 2,771,562
0.18%
7,650,200 7,664,305
0.92%
3,386,800 3,418,066
0.86%
1,532A56
0.77%
2,792,785
0.18%
7,678,436
0.92%
3,449,621
0.86%
1,545,601
0.77%
2,814,171
0.18%
7,692,593
0.92%
3,481,467
1996
0.62%
1,555,157
0.62%
2,831,483
0.15%
7,704,257
0.67%
3,504,910
1997
0.62%
1,564,773
0.62%
2,848,902
0.15%
7,715,938
0.67%
3,528,510
1998
0.62%
1,574,448
0.62%
2,866A28
0.15%
7,727,637
0.67%
3,552,270
Table A2.14
State
CT
MA
NY
NJ
State
CT
MA
NY
NJ
State
CT
MA
NY
NJ
Ann. Growth
Tot. Empl.
Ann. Growth
Tot. Empi.
Ann. Growth
Tot. Empi.
Ann. Growth
Tot. Empl.
2001
1.17%
1,711,232
1.28%
3,118,786
0.89%
8,300,677
1.16%
3,864,814
2002
1.17%
1,731,239
1.28%
3,158,750
0.89%
8,374,255
1.16%
3,909,836
2003
1.17%
1,751,481
1.28%
3,199,227
0.89%
8,448,486
1.16%
3,955,383
2004
1.17%
1,771,958
1.28%
3,240,221
0.89%
8,523,376
1.16%
4,001,460
2005
1.17%
1,792,676
1.28%
3,281,742
0.89%
8,598,928
1.16%
4,048,074
2006
1.17%
1,813,635
1.28%
3,323,794
0.89%
8,675,151
1.16%
4,095,230
2007
1.17%
1,834,840
1.28%
3,366,385
0.89%
8,752,049
1.16%
4,142,936
2008
1.17%
1,856,292
1.28%
3,409,522
0.89%
8,829,629
1.16%
4,191,198
2009
1.17%
1,877,995
1.28%
3A53,211
0.89%
8,907,897
1.16%
4,240,022
2010
1.17%
1,899,952
1.28%
3,497,461
0.89%
8,986,858
1.16%
4,289,415
Ann. Growth
Tot. Empl.
Ann. Growtt
Tot. Empl.
Ann. Growt[
Tot. EmpI.
Ann. Growth
Tot. Empl.
2001
1.92%
1,828,544
2.03%
3,332,634
1.64%
8,872,711
1.91%
4,129,623
2002
1.92%
1,863,637
2.03%
3,400,333
1.64%
9,017,906
1.91%
4,208,702
2003
1.92%
1,899,403
2.03%
3,469,408
1.64%
9,165,477
1.91%
4,289,295
2004
1.92%
1,935,856
2.03%
3,539,885
1.64%
9,315,462
1.91%
4,371,432
2005
1.92%
1,973,008
2.03%
3,611,795
1.64%
9,467,903
1.91%
4,455,141
2006
1.92%
2,010,874
2.03%
3,685,165
1.64%
9,622,837
1.91%
4,540,454
2007
1.92%
2,049,466
2.03%
3,760,025
1.64%
9,780,307
1.91%
4,627,400
2008
1.92%
2,088,799
2.03%
3,836,406
1.64%
9,940,354
1.91%
4,716,011
2009
1.92%
2,128,886
2.03%
3,914,339
1.64%
10,103,020
1.91%
4,806,318
2010
1.92%
2,169,743
2.03%
3,993,855
1.64%
10,268,348
1.91%
4,898,356
2001
0.42%
Ann. Growt[
Tot. Empi. 1,600,659
Ann. Growtt
0.53%
Tot. Empl. 2,917,224
Ann. Growth
0.14%
Tot. Empl. 7,761,662
Ann. Growth
0.41%
Tot. Empl. 3,615,208
2002
0.42%
1,607,368
0.53%
2,932,726
0.14%
7,772,250
0.41%
3,630,209
2003
0.42%
1,614,106
0.53%
2,948,310
0.14%
7,782,853
0.41%
3,645,271
2004
0.42%
1,620,872
0.53%
2,963,978
0.14%
7,793,471
0.41%
3,660,396
2005
0.42%
1,627,666
0.53%
2,979,728
0.14%
7,804,103
0.41%
3,675,584
2006
0.42%
1,634,489
0.53%
2,995,563
0.14%
7,814,749
0.41%
3,690,834
2007
0.42%
1,641,340
0.53%
3,011,481
0.14%
7,825,410
0.41%
3,706,148
2008
0.42%
1,648,220
0.53%
3,027,A84
0.14%
7,836,085
0.41%
3,721,526
2009
0.42%
1,655,129
0.53%
3,043,572
0.14%
7,846,775
0.41%
3,736,967
2010
0.42%
1,662,067
0.53%
3,059,746
0.14%
7,857A80
0.41%
3,752,473
Cro
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
1969
5-50
-
(D
UL
/0
T"
Employment
+
-
>
2006
2007 2008
2009 2010
2005
2004 -+
2003
1994
1995
1996
1997
1998
1999
2000
2001
2002
1993 --
1989
1990 19913
1992 -
1988
1987
1986
1983
1984
1985
1982 -
1980
1981U
1978 1979
1977
1976 -
1975
1973
1974
1969
1970
1971
1972
1
Employment
08
*
rn
~I1
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
II
-4-
~
~.
$a
T~
22o
I
U
U
U
U
U
U
U
U
/
U
U
U
U
U
*
*
U
*
II
*
11
II
II
II
II
II
I
I
U
U
U
U!
U
U
U
01
o o
Employment
R
(I~
O
1991
1992
1993
1994-1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
1989
1990
19691970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
I
I
*
I
C
0
I
a
(D
0
U
U
U
U
U
U
Employment
U!
U
U
*
*
U
U
U
1333C
(I~
(I~
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
Cl
C A
+
--
-
0
a
(-4D
r
I
U
U
U
U
U
U
U
U
*
i
U
U
U
I'
*
I'
*
I
*
.
II
I'
I'
I'
/
U
U
U
U
U
U
U
U
U
cn
Employment
average seasonal snowfall total of 83.9 inches. The pessimistic model assumes that
snowfall will remain at its current level of approximately 45 inches per season. Nominal
mortgage rates are excluded from the forecast since the investment model is not being
used. For purposes of calculating nominal prices and rents, a 4% annual inflation rate
could be assumed.
Skier Market
Figures Al. 14a-c show the projected skier visits for the different employment
growth scenarios. Figure Al. 14a shows the results of the expected employment growth
with a return to the long term average snowfall. As shown in the figure, skier visits are
expected to decline through the '93-'94 season at which point they will bottom out and
start a steady upward trend. Assuming snowfall dropped off and employment was as
expected, skier visit increases would experience a slight blip before continuing to rise.
Snowfall will have a smaller and smaller impact as time goes forward. Areas will rely
upon snowmaking equipment to bring skiers to their resorts. The most critical variable
going forward is the regional employment. According to this first model, skier visits will
not reach their peak 1987 level again until the turn of the century.
If we look at the optimistic set of assumptions as portrayed in Figure Al. 14b, we
see a slight dip in visits in the '93-94 season and then a sharp, steady increase in visits as
employment continues to grow. In this case, skier visits would match their 1987 level by
the '96-'97 season.
Finally, the effect of the pessimistic employment and snowfall assumptions are
charted in Figure A1.14c. The skier visits continue to drop through the '93-'94 season and
stage only a modest increase over the next sixteen years. They never recover to reach the
levels of the late-eighties. While such a scenario seems unlikely, it is not impossible. If
employment grows at a sluggish pace due to a decreased growth in the working
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
S1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
0I
Q
J>
a0.
(D
0L
DI
0(
DI
DI
Skier Visits
ii 0
2007
2008
2009
2010
2006 --.
2005-->
2004 --
2003 - -
2002 --
2001 -
2000 -
1997 -1998 -1999
1992 1993 1994 1993
1995 --
1989 -0-C1990 --
1988 -
1987 -
1974 1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1972 -1973 -
1971
-
il
a
0.
?"
i
-
x
I
ll
I
Skier Visits
l
I
io
O
*
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
c,)
(D
0n
a
o c 05
81
8
Skier Visits
population (aging baby boomers), and snowfall stays low, we may see a very sluggish
improvement or perhaps no improvement in skier visits over the next decade and a half
It is important to note that these are trends based on trends in employment. There
will be some up and down years between now and 2010 as employment undergoes
unexpected surges and drops brought on by booms and recessions. The trend, however,
should hold true as long as the long term trend in employment holds true. These
projections should obviously be reevaluated as the employment projections change. It is
interesting to note that the trend line for the expected scenario has a similar slope to the
historical trend while the optimistic forecast has a steeper slope and the pessimistic
forecast has a more gradual slope.
Rental Market
Condominium
The plots of forecasted real condominium rent indices are shown in Figures
Al. 15a-c. According to the basic model shown in Figure Al.15a, real
condominium rents will decline over the next several years before bottoming out in
the '94-'95 season. They will then undergo steady exponential growth. They will
reach the levels of the mid- to late-eighties around 2003 and will continue to
increase in real terms as long as visits and employment are increasing. The
continued decline in rents through 1995 is due to the lagged effect of changing
employment on skier visits and the lagged effect of skier visits on rent. This
model, as well as all the others assumes that employment will start increasing in the
current year (as predicted by the economists). If employment continues to drop,
the turnaround in skier visits and rents will continue to be delayed.
If condominium owners change their rent setting practices as discussed
previously and set seasonal rents based on expected visits in a given season, they
could avoid this lagged effect on rent. They would be able to cash in on the
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
-<1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
(D
c
-0-
(D
(D
OL
-0
(
aL
U
U
Real Condo Rent
68
U
:r
0
0
0
0
0-e0
1-11
2009 -2010 -D
a.
-
200 7 -2008 -+-4-
-o
I
!
2006 -
2005
2004
2003
1994
1995
1996
1997
1998
1999U
2000
2001
2002
1993
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
Real Condo Rent Index
J11992-
-
2010
2009
2006
2007 2008
2005
2003 2004 -
2002 --
2001
2000 -
1996
1997
1998
1995 -!-
1993 -*1994
1990 1991 - -
1989 - -
1988
1987
1985
1986
1984
1982
1983 -0-
1979
1980 1981
1978
1977
1976
1975
1974
1973
1971
1972
o
I
8_
I
0.
(DU
0.
"
8_
Real Condo Rent Index
_8
I
I
0
0
0.
increase in skier visits in the year in which they occur, rather than waiting until
subsequent years.
If the New England and mid-Atlantic economies grow faster than expected
over the next 17 years, the results may look like those shown in Figure Al. 15b.
The rents drop only slightly in the next year and will then begin a drastic rebound.
By 1999 the rent will have exceeded the levels of the eighties and will continue to
rise exponentially. As above, there is still a lagged effect on rents by employment
and skier visits, but the rent levels will be corrected quickly if these optimistic
employment figures can be attained.
The other alternative, of course, is that we continue the sluggish recovery
of the current economic recession for the next 17 years. In this case, the rents will
look more like those shown in Figure Al. 15c. Real rents will continue to drop
through about 1999 at which point they will begin a modest improvement. By
2010, however, they still will not have increased to the levels of the early seventies.
This would clearly spell bad news to investors in ski resort real estate. If the
regional economy does not start to show signs of a solid recovery soon investors
relying upon rental income should consider liquidating their assets now.
Hotel
The forecasts for the hotel market are shown in Figures Al. 16a-c. As
expected, hotel rates move directly with skier visits, which in turn move with
employment. As discussed previously, hotel operators do a better job of
estimating skier visits in a current season and setting the rates accordingly.
In the basic growth model shown in Figure Al. 16a, hotel rents will
continue to decline slightly and remain flat until the '93-'94 season. They will then
begin a steady increase with increasing skier visits. As shown in the figure, they
will start to follow an upward trend at a rate similar to that experienced from the
late-seventies through the mid-eighties (approximately the same slope) before the
C
-
-
2009
2010
-
2008 -+
2007 -
2006 --5
2005 -
2004
1996
1997
1998
1999
2000
2001
2002
2003 -
1995a
1988
1989
1990
19
Q 1991
1992
1993
1994
1986
1987
19711972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
0
F8
I
(D
8
*
F88
Real Hotel Rent
I
I
-
-21
2009 2010
2008 -4
2006
2007
(D
#(D
(D
0
2005
;j
(D
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
018
Real Hotel Rent Index
0
2007 --a
2008 -2009 -2010
2005 -2006 --
2003
2004 --
2002
2001 -
2000-
1999
1995
1996
1997
1998 -
1994
1993
1992 -
(D)
0
1989
1990
0.
U
-- I----
1982
1983
1984
1985 -1986 - 1987 1988
1981 -
19 71
1972
1973
1974
1975
1976
1977
1978
1979
1980
Real Hotel Rent Index
market was jolted out of equilibrium with a sudden increase in the late-eighties and
the subsequent drop in the early-nineties. Assuming employment grows at a steady
moderate pace, the hotel rates will follow with a steady upward trend. The erratic
price movements at the turn of the last decade can be thought of as a blip in the
upward trend of hotel rates, similar to the blip that is just shown to be corrected at
the beginning of the seventies. The market strives to reach long term equilibrium
over time.
If the economy grows faster than expected, the results may look similar to
those shown in Figure Al. 16b, assuming an optimistic view of the future. Hotel
rents will immediately begin to increase with the increasing employment. They will
increase at a rate greater than that experienced from the late-seventies to mideighties.
Finally, if the economy grows at a slow rate corresponding to the
pessimistic outlook, the results may look like those shown in Figure Al. 16c. In
this scenario, the rates would continue to drop through the '95-'96 season and
would experience very slow growth for the next 15 years, never regaining the
levels of the late-eighties. This growth of rents would be slower than that
experienced in the late-seventies to mid-eighties, indicated by a flatter trend line.
This would indicate very bad news for hotel owners and operators.
Asset Market
The asset market, unlike the skier and rental markets, is poised for immediate
recovery as soon as employment starts to turn around. Two separate price models have
been included in the forecast, representing the two consumer models discussed in the asset
market section. While the results of these models differ slightly over the long term, they
both show immediate improvements in the market as soon as employment starts to
increase. The investment models were not included for the purposes of forecasting since
they were found to be flawed as previously discussed. The incorrect signs on the models
led to ridiculous results.
Figure Al. 17a shows the prediction for the first consumer market model (including
skier visits) under the expected economic conditions. An immediate improvement in the
real price index is expected in the first year and then a gradual increase in the index will
occur over the next decade and a half. The index will remain relatively low throughout
this period, however, as the stock continues to increase and offset price increases. By
2010 the index should reach the level it was at during the recession of the late seventies,
well below the levels of the mid- to late-eighties.
The other consumer model projection is shown in Figure Al. 18a for the same
economic estimates. This model shows a slightly greater increase in the index. It begins
with a robust increase and tapers off over time and by 2010 reaches the 1989 price level,
still less than the peak of the late-eighties. Unlike the other model, this model is
independent of skier visits and bases its demand on the employment in the regional
economy. The first model was a function of employment and snowfall, and snowfall was
not increasing significantly.
For a more optimistic economic outlook the pictures look somewhat different as
shown in Figures Al.1 7b & Al. 18b for the skier and employment models respectively.
For the first model, the prices begin to rebound immediately and grow linearly through
2010. By 2005 the price index will have increased to its highest historical level and will
continue to increase. In the second model, the growth starts with a steeper rate and
begins to taper off over time. By 2004 the price index will set a new record level and
continue to increase. The models are similar in their estimates of the future trend of the
index.
Finally, in the more pessimistic economic model the situation looks substantially
worse. Figures Al.17c & Al.18c show the two models again with the skier visits model
first. Given the low snowfall accumulations, the lower skier visits only aggravate the poor
--
2010 -D
2009 --
2007 -D
2008 -
2006
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
1995
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1984 -
1976
1977
1978
1979
1980
1981
1982
1983
1975 -
19711972
1973
1974
0
:F
5
-
Real Price Index
~
-
2010
2007
2008
2009
-
T
2005 2006
2004 --
2003 ---
2002 --
2001
1999-2000-
1998 -
1993
1994
1995
1996
1997
1992
1989
1990
1991
1985
1986
1987
1988
1984
1977
1978
1979
1980
1981
1982 -0
1983 -
1975 1976 -
1974 -
1971
1972 -1973
C0'
I
(71
(
(D)
0-
V
|
|
(1
N)0
(1
Real Price Index
66
I
(0
I
(0
a
'
0
0
0
0
0
0
*
~1991
2008
2009
2010
-0-
-
2007
2006
2005
2004
2003
2002
1998
1999 -2000 -I
2001
1996
1997
1995
1994
1992 1993
1988 1989
1990
1986 1987 -
1985 -
1984 -
1983 --
1978
1979
1980
1981
1982 --
1977 -
1976
1975
1972
1973
1974
1971
0
- l
0
01
0
-
-
X
I
(D)
0'n
Real Price Index
001
1
CA
C
J
C\
0
0
_
0
-
OA
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
1988
1971
1972 1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
0
0
I
0
(D
(L
xL
cT
=.
0.
0-
Real Price Index
101
n
cI )
-
-
2010
2004
2005
2006
2007
2008
2009
--
-
2003 -.-o
2002
2001
2000
1996
1997
1998
1995
O-q1991n
1992
1993
1994 -
1990
1988
1989
1984
1985 -1986
1987 -
1983 - -
1976
1977
1978 1979
1980
1981
1982
1974
1975
o
1971
1972 -1973
(D5
[
-CCnN)
to'
('0
m
Real Price Index
-
0
0
0
0
C
0
(I
-
-
-
2009 2010
2007 -D
2008 - -
2006
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
-
-
1995 -
1993
1994
1986
1987
1988
1989
1990
1991U
1992
1985
1982
1983
1984
1981
1980
1978
1979
1972
1973
1974
1975
1976
1977
1971
0
I
0
01
!T
X
C
?
0)
0
(
-CCnN)
Real Price Index
E01
m
N
0
CA)
0
economic conditions in the first model. The price index continues to fall and then
stagnates indefinitely at a record level below the 1991 level. The model that doesn't
consider skier visits is less affected, however, the performance is less than stellar. The
index inches up slightly, but by 2010 it has still not reached the mediocre levels of 1990.
Which model should be used? Intuitively, prices of ski resort real estate should be
a function of the number of skier visits to the area and not simply a function of the
regional population. If the number of skiers drops off, the market for ski real estate will
fall off and prices will drop. On the other hand the importance of the regional employment
should not be discounted. If people do not have jobs they certainly will not be able to
afford second homes although they may be able to afford a few trips to ski country.
In all likelihood a combination of the two models is the best estimate of the trend
in the price index. For the purposes of estimating completions, which is a function of the
price indices in prior years, the average of these two price indices was used.
The other aspect of the asset market in addition to prices is the annual number of
completions and the corresponding stock of condominiums. As mentioned, the forecast of
completions used the average of the price indices from the two consumer models. The
predicted completions are shown in Figures Al. 19a-c. For the expected economic
outlook the completion data is shown in Figure Al. 19a. Completions are expected to
jump above zero in 1993 to five (although the author was unaware of any construction at
the time of writing). They will increase until 1994 after which they will decrease until
1996. After 1996, completions are expected to edge upward in each year reaching 240 by
the year 2010.
For the optimistic outlook shown in Figure Al. 19b, there will be the same five
completions in 1993 followed by an increase in 1994, a stagnation until 1996 and then a
steady annual increase to 454 per year by 2010.
104
1990
g
-~
a
I
I
2009
2010
2008
2006
2007
2005
2004
2003
2000
2001 -2002-
1999 - 1998
199839
1992
1991
1988 -
U
3
0
cn
C
0
=
VM
0
0
0
0
3
I
1986
1987
I
0.
I
1985
1984
U"
o
I
0
0
-0.
-
l
1982
1981
1974
1975
1976
1977
1978 1979
1980
1973
1972
Annual Completions
So I
-
2001
2008 -+
2002
2003
2004
2005U
2006
2007
2008
2009
2010
2000
1999
1998
199
1997
1994 -
1991
1992I
19950
1988
1986
19873-
1984
1981
1984
1982 -19851
1980
1974
1975
1976
1977
1978 1979
/
"
\+~T
\
1989
1985
00
3
0
a
0
(D
20
20 0
Annual Completions
901
(
0
3-n
0
3
0
0.
0
-
0
I
0
I
(D4
|
|
-
0
2002 -
2007 2010
2008
2006
2005
20043
2003
-
-
2002 --
2001 --
2000
1997
1995
1994-
0U
0
19930
199
1992
(
0
0
3
1988
1987
1986
1985
0
"3
I
19853
1984
I
0
l
1982
1981
1980
1973
1974
1975
1976
1977
1978 *
1979
19 72
Annual Completions
LOI
Finally, in the pessimistic outlook shown in Figure Al. 19c, there is a spurt of
construction activity between now and 1995 which dies off and then modest annual
construction gains are made starting in 1998 edging up to 53 units per year in 2010,
substantially smaller figures than those estimated in the first two scenarios.
It should be noted that these completions projections are mathematical predictions
based on past building trends and do not take account of physical constraints associated
with the topography. It may be impossible to sustain completion rates like those estimated
in the optimistic scenario for any length of time. These constraints will in turn place
external restrictions on the market and will artificially drive up the price and rent indices
by restricting supply.
Summary
In summary, the future projections of skier visits, condo rent, hotel rent, condo
prices and completions are entirely dependent upon the accuracy of the projected input,
primarily employment and to a lesser degree snowfall. Assuming the employment growth
projections are nearly correct, skier visit should drop slightly over the next year or two
before rebounding to previous levels. Condo rent will also continue a decline over the
next couple of years before staging a moderate increase as skier visits trend upward.
Hotel rent will continue to drop over the next couple of years before reestablishing its
moderate long term growth trend. Prices will begin a weak to moderate recovery as soon
as employment starts to increase, but will take a long time to reach the record levels set in
the mid- to late-eighties. Finally, completions will trend upward over the next decade and
a half as the increasing price indices warrant the new construction, assuming there are no
physical constraints to building.
The severity of the current economic recession on the ski resort real estate
industry, however, can not be overstated. The industry has been hit hard and has seen
substantial drops in value that will not be recovered for many years to come. Properties
108
are typically worth about two-thirds of what they traded for in the mid- to late-eighties.
This means that a lot of people lost a lot of money and it will take a long time before those
losses can be recovered.
109
...........
Summary of Results
What are the results of this study? We have been able to develop predictive
models for the behavior of skier visits to the State of Vermont and the cycle of real estate
rents, prices and stock at a single New England ski resort. These models are only
dependent upon the exogenous variables of employment in Connecticut, Massachusetts,
New York and New Jersey, and the natural snowfall in Sherburne, Vermont, with the
primary emphasis on employment.
How were these models created and what do they mean? The first step in
developing a model is to understand exactly what is being modeled. In this case, the
model was looking at Type II New England ski resorts which characterize all the major
resorts in New England. They are dependent upon a market which is within driving
distance, and depend primarily upon weekend skier traffic and to a lesser extent mid-week
traffic (with the obvious exception of holiday weeks which they are very dependent upon).
They are also dependent upon artificial snowmaking capabilities in order to provide quality
skiing, since seasonal snowfall in the East has been declining over the years.
In order to keep the task of manageable size and scope, only one New England
resort was studied and only condominium ownership (as opposed to single-family) was
considered. The use of similar condominiums made data collection much easier since
fewer sales were required in a given year to establish a price index. The only form of
commercial ownership studied was the hotel market. This was considered for comparison
with the condo rental market.
According to all the literature studied, the market for second homes and ski resort
real estate is expected to blossom over the course of the next decade as the baby-boomers
110
and dual income earners age into the bracket of second home buyers. As this occurs, the
skier market will increase some, but will not necessarily follow the growth in second
homes since the typical skier is of a younger age than the typical second home buyer. As
the baby boomers age, they will pass out of the typical skier bracket. To make up for this,
the ski industry will have to be promoting skiing as a sport for all ages. While the skier
market was not profiled by age in this study, the characteristics have been changing over
the more recent years to reflect a greater influence of employment. This may indicate the
age of the average skier is increasing, however, a more comprehensive study should
consider the effect of different age brackets more precisely.
The many factors that affect ski resort real estate prices were outlined, including
the effects of the national, regional and local economies and real estate cycles. The basic
conclusion reached is that a particular ski area's prosperity is a function of the prosperity
in the market being served. In the case of a large Type I resort this would be the national
economy, in the case of a Type II New England resort, it would be the economy of the
metropolitan regions within driving distance.
Other factors affecting value of course are the stock of real estate (supply), the
number of skier visits (demand), taxes (both property and income), financing arrangements
(although they were found to have a small effect), and the less tangible factors such as
resort characteristics, site/unit characteristics, and land-use regulations. All these factors
combine to create a market value.
In order to predict this market value we needed to start with historical data.
Condominium sales data was gathered and used to create annual price indices for each
year, and condominium and hotel rental data was used to develop annual rent indices for
each year. These indices automatically took care of the intangible factors. Taxes and
financing were found to be of little significance in this case, which left the remaining
factors of regional economy, skier visits and stock. The only exogenous variable
111
necessary then became employment and snowfall. These factors were used to develop the
predictive equations outlined in the study.
What does all this mean? Based on the models, skier visits were found to be a
function of the regional employment and snowfall. Initially, snowfall was the primary
factor driving skier visits, however, this has changed over the years and employment is
now the dominant determinant of visits. This is presumed to be the combined result of
increased snowmaking capabilities and the increasing cost of skiing.
Last year's skier visits then combine with the last year's rent and condo stock to
determine this year's condo rent. On the other hand, this year's skier visits and last year's
hotel rent combine to yield this year's hotel rent. It was found that hotel operators have
been more successful than condo owners at predicting the skier visits in the current year
and setting the rents accordingly. Instead condo owners continue to bench mark off last
year's performance and lose money in the process. This was believed to be the result of
the owner - agent dilemma.
Two basic models were developed to explain the asset price market. The first
model was the traditional investment model which valuates an asset based on its expected
return. This model was found to be completely inaccurate and established that the ski real
estate market is not an investment market. It is possible that people invest in ski real
estate for investment purposes, but it likely that many of the people doing this are not
sophisticated investors, but rather consumers who see a personal interest in owning the
property and expect some rental income on the side to help offset the cost of the property.
They do not evaluate the property to see if it is a self-sustaining business. While some
people may invest in ski real estate as a business, in general they will be priced out of the
market by a consumer who is less interested in the rental income stream.
The model that does predict prices well is the consumer model which is a function
of the annual change in regional employment, the condo stock two years ago, the price
index last year, and either the skier visits in the upcoming season or the total regional
112
employment this year. Both these models have good statistical fits and represent the two
potential buying groups. Skiers and non-skiers who are simply interested in a mountain
vacation home. For purposes of forecasting, the average of the two was used.
The last model that was created used the models built before it. Completions were
found to be a function of price and rent, which of course are a function of skier visits,
employment and each other. In particular, completions are a function of last year's condo
rent, last year's change in the price index from the previous year, and last year's change in
the rent index from the previous year. There was some degree of difficulty in getting a
good fit between completions and this data due to the cyclical nature of completions.
Some years there were no completions while some years had several hundred.
The effects of supply and demand on the price index can be seen in Figure Al. 12
as shown on page 74. This plot shows how the price index moves with the rent index and
completions. In this case, the rent index is assumed to be a measure of demand (function
of skier visits) while the completions are a measure of stock. As demand increases and
supply decreases, prices go up. As demand decreases and supply increases, prices go
down. When demand and supply move together a balance is reached depending on which
moves more.
The forecasts that were run with these models were based on expected future
employment growth, and indicate that the road to recovery of real estate values will be a
long, slow, bumpy one. Skier visits are expected to drop slightly over the next couple of
years before a solid recovery in visits will be under way. Condo and hotel rents will
follow a similar trend with an initial drop followed by slow, steady growth. Condo prices
will begin a gradual increase as soon as employment starts rising, but will not reach mideighties levels for a long time to come. Completions will begin an immediate recovery
followed by a couple slow years and will then experience steady growth, assuming the
area can sustain the growth and developers learned enough of a lesson in the eighties not
to charge ahead full steam at the first hint of a recovery. The industry has undergone a
113
structural change over the past four years as the real estate cycle has been jolted out of
long term equilibrium. Assuming the economy can stabilize and job growth can begin a
slow steady increase, real estate values will settle into a new long term trend, but it will
take a long time for values to get back to the real levels of the booming eighties.
Applicability to other New England Ski Resorts
Is this study applicable to other New England resorts? The simple answer is yes.
New England ski resorts are very similar in their character and nature. The first aspect of
the study, skier visit prediction would be identical for any mountain in southern Vermont
that was served by the New York and New Jersey markets in addition to Connecticut and
Massachusetts. The data on skier visits was taken for the whole state of Vermont with no
particular connection to Killington. For a resort in New Hampshire or Maine, the more
appropriate employment region would probably include Connecticut, Massachusetts, New
Hampshire and Maine. The trends should still hold true, however, that visits are
predominantly a function of regional employment.
In terms of the rental and asset market data the obvious difference is that this study
was conducted at five specific condominium complexes and four hotels at Killington. The
trends, however, should still hold true and the variables used to estimate rents, prices and
completions should also be the determining factors at other New England resorts. Since
Killington is sharing the same regional market with Stratton and the other southern
Vermont resorts, the forecasts should hold for these resorts as well, with dropping visits
and rents over the next couple years followed by steady growth in each, and slowly
increasing prices and completions. Similar trends should hold for New Hampshire and
Maine resorts since Connecticut and Massachusetts employment are forecast to grow at
similar rates to New York and New Jersey.
114
Applicability to Recreational Real Estate in General
Is this study applicable to recreational real estate in general? Once again the
simple answer is yes. Ski resorts are New England's favorite winter resorts and are
striving to compete with New England's favorite summer resorts - golf and water related
resorts. People desire recreation in the summer as well as the winter, and in all likelihood
consider the same economic factors before renting or buying a second home, be it a
summer condo (or house) or a winter condo (or house). The same economic factors that
determine ski condo prices should determine beach condo prices. Obviously the demand
function would be different than skier visits, but the effect of employment on prices should
be similar.
The Future of the Second Home Market
What does all this mean for the second home market? Based on the predictions of
the experts discussed previously in the study, there should be a surge in the demand for
second homes over the next decade as baby boomers age into the traditional second home
buying bracket. The models that have been created do not forecast such a surge over the
next decade, rather they predict a steady, moderate growth in demand. The models,
however, do not incorporate demographic data which would support such a surge. The
models are based entirely upon historical information, which may not adequately represent
the effects of a surge in a particular age group. It would, however, model the effects of a
surge in employment.
A more comprehensive study should break down the general employment figure by
age bracket to determine more accurately who buys real estate and what the effects of a
surge in a given bracket would be. Due to time constraints, however, this level of study
was simply not feasible.
Based on current research, demographic trends, economic forecasts and the results
of this study, the second home market may be poised for a boom, but it is atleast due for a
115
recovery. Depending upon which employment forecasts are accurate and if baby boomers
really will rush to buy second homes within three hours drive of their primary homes,
prices should atleast begin to climb having hit the bottom of their slide. Although rents
will continue to fall slightly over the next couple years before bottoming out, they will not
decline significantly and will be followed by a steady growth trend as the baby boomers hit
their stride. It can not be emphasized enough, however, that this growth will be slow.
The high rolling days of the eighties are gone and current values are substantially lower
than were. Even if baby boomers do rush to buy homes, it will take a long time to reach
the real price levels of the last decade.
116
.:~~:::
~:~~x
.ox4~*'"~
n:~:~:~n
."...,.
fl:MIM
...
t~******
--- - -*---
117
Table A3.1
Data Summary
Real
Year
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
Price index
1
Condo
Real
Hotel
Real
130
124
128
133
149.43299
154.529412
166.117647
169.938144
192.185567
218.721649
239.556701
268.752577
322.235294
392.329897
391.371134
449.28866
423.470588
443.247059
444.188235
507.294118
523.858824
486.752941
419.364706
446.97337
408.53828
409.36937
400.12712
408.85327
390.44798
397.11626
381.82305
404.30149
419.70911
413.32991
423.56122
478.63401
559.34584
534.88644
592.92741
548.65715
554.05882
533.17607
580.93359
569.15037
507.48104
419.36471
37.6805556
36.025641
37.1752137
37.8418803
42.5811966
44.9401709
45.0470085
50.1452991
54.6581197
63.4358974
69.5854701
79.1196581
86.517094
93.9145299
101.205128
100.205128
114.252137
132.166667
154.350427
142.290598
146.521368
142.269231
140.452991
129.55542
118.69237
118.8937
113.84633
116.50347
113.5499
107.68813
112.66824
114.98449
121.72834
120.06241
124.69469
128.50865
133.89421
138.31692
132.24097
148.0274
165.20833
185.2727
162.94569
159.18924
148.32769
140.45299
Price Index Rent Index C Rent Ind. Rent Index I H Rent Ind. I Skier Visits(VT) I Snowfall (in) Condo Stock
3.60406091
0.80346994 2.76253589
0.85605499 2.82041318
0.92458705 2.95701264
0.98464052 2.96226598
1.01962846 2.78973489
1.09247413 2.76034388
1.11803534 2.67274442
1.14347723 2.56920518
1.33774125 2.81421122
1.49764332 2.87385611
1.72248373 2.97196463
2.0284035 3.19681795
1.82223013 2.70665981
2.16457789 3.08604478
2.23225282 3.05081714
2.42564175 3.20112573
2.34284724 3.03544077
2.67503429 3.34379286
3.1929036
2.66000349
2.45916672 2.81614253
2.3254203
2.14036925
1.81999753
1.8975011
2.00544693 2.00544693
2,250,000
2,400,000
2,650,000
2,650,000
2,300,000
1,650,000
2,800,000
2,600,000
3,000,000
3,600,000
3,200,000
2,100,000
3,100,000
4,000,000
3,000,000
4,150,000
3,850,000
4,460,000
5,200,000
4,850,000
4,500,000
4,600,000
4,100,000
4,300,000
74.3
131.5
112.5
96.6
71
96.8
102
108.3
127.5
96.7
41.6
83.4
93.7
81.6
79.6
79.2
76.2
72.2
82.1
47.8
78.2
48.2
48.8
58
98
98
98
194
194
194
194
194
220
240
252
252
660
744
1,079
1,186
1,318
1,374
1,477
1,477
1,477
1,477
1,477
Table A3.1
Emplo ment Data
Total
Year
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
Connecticut
1,194,100
1,197,500
1,164,300
1,190,400
1,238,700
1,264,000
1,223,400
1,239,700
1,282,300
1,346,100
1,398,000
1,426,800
1,438,700
1,429,800
1,446,500
1,520,500
1,562,300
1,604,200
1,644,700
1,674,900
1,674,100
1,632,900
1,557,800
1,506,500
Massachusetts
2,249,400
2,243,500
2,211,400
2,251,700
2,333,500
2,353,700
2,273,100
2,323,500
2,416,000
2,526,300
2,603,500
2,652,200
2,668,300
2,638,000
2,692,500
2,851,800
2,926,000
2,984,800
3,061,800
3,126,200
3,103,400
2,979,000
2,817,000
2,750,500
New York
7,182,000
7,156,400
7,011,400
7,038,500
7,132,200
7,078,000
6,829,900
6,789,500
6,857,600
7,044,500
7,179,400
7,207,100
7,287,300
7,254,600
7,313,300
7,572,300
7,750,800
7,904,400
8,059,400
8,186,900
8,246,800
8,213,000
7,885,800
7,650,200
Change in IChange
New Jersey i Emlo ment Employment IEmploymentj
2,569,600
2,606,200
2,607,600
2,672,500
2,759,700
2,783,000
2,699,900
2,753,700
2,836,900
2,961,900
3,027,200
3,060,400
3,098,900
3,092,800
3,165,100
3,329,200
3,414,100
3,490,500
3,581,600
3,659,500
3,689,800
3,642,300
3,493,100
3,386,800
13,195,100
13,203,600
12,994,700
13,153,100
13,464,100
13,478,700
13,026,300
13,106,400
13,392,800
13,878,800
14,208,100
14,346,500
14,493,200
14,415,200
14,617,400
15,273,800
15,653,200
15,983,900
16,347,500
16,647,500
16,714,100
16,467,200
15,753,700
15,294,000
8,500
-208,900
158,400
311,000
14,600
-452,400
80,100
286,400
486,000
329,300
138,400
146,700
-78,000
202,200
656,400
379,400
330,700
363,600
300,000
66,600
-246,900
-713,500
-459,700
0.06%
-1.58%
1.22%
2.36%
0.11%
-3.36%
0.61%
2.19%
3.63%
2.37%
0.97%
1.02%
-0.54%
1.40%
4.49%
2.48%
2.11%
2.27%
1.84%
0.40%
-1.48%
-4.33%
-2.92%
Table A3.1
Economic Data
Year
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
Nominal
Mortga e Rates
7.67%
8.22%
7.56%
7.40%
7.80%
8.76%
8.92%
8.87%
8.82%
9.37%
10.59%
12.46%
14.39%
14.73%
12.26%
11.99%
11.17%
9.79%
8.95%
8.98%
9.81%
9.74%
9.01%
7.98%
CPI
39.4
41.3
43.1
44.4
47.2
51.9
56.2
59.4
63.2
67.5
74
82.3
90.1
95.6
99.6
103.9
107.6
109.6
113.6
118.3
124
130.7
136.2
142
.
Inflation
4.82%
4.36%
3.02%
6.31%
9.96%
8.29%
5.69%
6.40%
6.80%
9.63%
11.22%
9.48%
6.10%
4.18%
4.32%
3.56%
1.86%
3.65%
4.14%
4.82%
5.40%
4.21%
4.26%
Real
Mortgage Rates
3.40%
3.20%
4.38%
1.49%
-1.20%
0.63%
3.18%
2.42%
2.57%
0.96%
1.24%
4.91%
8.63%
8.08%
7.67%
7.61%
7.93%
5.30%
4.84%
4.99%
4.34%
4.80%
3.72%
Avg. Real
Mortgage Rates
3.40%
2.97%
3.33%
3.24%
2.33%
0.74%
0.89%
2.03%
3.07%
2.98%
3.24%
4.28%
5.80%
5.67%
7.12%
7.15%
6.54%
5.93%
5.76%
5.61%
4.95%
4.20%
3.36%
Table A3.2
Sales Data
Identifier
P03
P05
P08
P01
C03
C04
C05
C06
P02
EA6
EA7
EA4
EB7
EA8
EB2
EB4
EB8
EB6
EB3
P14
P13
P16
P15
EBl
EC3
ECl
EA5
EC2
EA3
EB5
EC6
P09
EC4
EC8
EC5
EE 1
EE3
EE4
EE6
C08
ED6
EE8
EE7
EE2
EE5
ED5
P06
ED7
ED3
ED1
P1l
ED4
Sale Price
Date
$24,000
10/3/67
$24,000
11/30/67
5/9/69
$25,600
8/6/69
$26,500
12/29/69
$39,500
1/5/70
$37,000
$40,000
2/18/70
$35,000
2/24/70
7/24/70
$30,000
12/18/70
$33,900
12/18/70
$22,555
12/23/70
$37,400
12/23/70
$22,900
12/24/70
$35,400
12/24/70
$34,900
$34,400
12/24/70
12/24/70
$34,400
1/24/71
$34,400
$22,900
1/30/71
2/1/71
$36,000
3/12/71
$33,840
7/21/71
$31,000
$29,000
9/23/71
10/12/71
$23,500
10/15/71
$23,500
$24,300
10/16/71
$23,100
10/20/71
$38,400
10/20/71
10/22/71
$23,100
10/22/71
$23,100
$37,900
10/22/71
$31,000
10/29/71
11/11/71
$37,400
$38,600
11/29/71
12/10/71
$23,900
$24,300
12/10/71
12/10/71
$23,900
$37,900
12/13/71
$37,900
12/13/71
$43,000
12/17/71
$37,900
12/18/71
$37,900
12/18/71
$23,900
12/20/71
12/21/71
$38,400
12/23/71
$23,900
12/31/71
$23,900
1/3/72
$32,500
$24,300
1/7/72
1/9/72
$23,900
1/10/72
$23,900
1/15/72
$45,000
1/17/72
$37,400
121
Table A3.2
ED8
ED2
P12
EC6
C10
EA2
C09
P20
P22
P18
C07
P24
P17
P13
P19
P21
P23
H2C
H2D
H3D
H2A
P06
H2B
H4A
H3A
H3C
P14
WC2
WB1
WC4
WB3
WB8
WAl
WB4
WB6
H4B
WA6
WA4
WB5
WC1
WC3
WC6
P1O
H4D
H4C
H3B
H6C
H5B
H5C
WB2
WA7
H5D
WE3
H6D
WA2
2/1/72
3/6/72
6/16/72
9/22/72
9/22/72
10/10/72
10/13/72
12/1/72
12/26/72
1/8/73
1/10/73
2/20/73
3/6/73
3/28/73
3/28/73
3/30/73
4/3/73
4/20/73
5/10/73
5/12/73
5/18/73
6/1/73
6/4/73
6/19/73
6/21/73
6/21/73
7/2/73
7/20/73
7/23/73
7/23/73
7/24/73
7/26/73
7/27/73
7/27/73
7/27/73
7/27/73
7/28/73
7/30/73
7/30/73
7/30/73
7/30/73
7/31/73
8/21/73
8/27/73
9/7/73
9/28/73
10/12/73
10/22/73
10/23/73
10/26/73
11/2/73
11/2/73
11/19/73
11/19/73
11/24/73
122
$38,900
$37,900
$43,500
$44,000
$43,000
$36,500
$43,000
$44,000
$44,000
$43,500
$43,000
$53,000
$42,650
$37,000
$43,000
$45,000
$41,300
$36,500
$36,100
$37,500
$37,500
$40,900
$36,500
$38,500
$37,000
$36,000
$40,000
$31,750
$31,500
$32,750
$32,500
$48,500
$40,500
$28,000
$29,250
$36,150
$41,000
$42,000
$32,500
$28,250
$28,000
$32,750
$33,000
$38,250
$36,750
$36,750
$36,750
$36,700
$36,250
$28,800
$61,300
$38,250
$32,750
$43,250
$42,300
Table A3.2
WF6
H6B
H5A
ED2
WD8
H6A
HIC
C03
HIA
WD7
WE5
HiB
P06
WF7
P16
H2D
WE6
WF4
WE8
WC5
WF3
WE4
P16
WF2
WF1
C06
C09
WF5
WEl
HID
H3D
WD6
H5B
WD1
WE2
WA5
EC4
P15
WA4
P1l
EE8
WC4
HIB
H2D
H2B
P02
C03
WD4
C04
WD5
WD2
H4D
WC7
WC8
EA3
11/26/73
11/28/73
12/28/73
5/6/74
6/3/74
7/3/74
9/20/74
9/27/74
10/11/74
10/15/74
10/31/74
11/5/74
12/3/74
12/23/74
2/13/75
7/16/75
7/24/75
11/17/75
12/8/75
12/12/75
12/12/75
12/15/75
12/15/75
1/13/76
2/4/76
3/19/76
5/5/76
6/30/76
7/15/76
9/17/76
9/20/76
10/15/76
10/30/76
12/1/76
12/1/76
12/27/76
1/3/77
1/10/77
1/28/77
2/17/77
3/16/77
3/25/77
6/3/77
6/23/77
7/8/77
9/2/77
9/9/77
9/16/77
9/16/77
9/17/77
9/22/77
9/27/77
10/31/77
11/2/77
11/4/77
123
$34,300
$37,200
$37,840
$43,000
$60,000
$38,250
$37,250
$52,000
$38,750
$63,250
$35,480
$37,250
$30,000
$53,275
$36,000
$48,500
$32,575
$37,260
$55,725
$33,075
$30,796
$30,795
$40,000
$34,722
$32,075
$44,000
$51,500
$33,075
$36,760
$45,000
$47,500
$46,935
$46,000
$45,935
$32,075
$48,075
$41,300
$40,500
$40,000
$54,000
$48,500
$33,750
$49,500
$40,000
$45,000
$44,500
$55,000
$47,575
$54,200
$48,075
$47,075
$49,500
$56,000
$56,725
$30,000
Table A3.2
WE6
WF6
EA8
EB6
WAl
WF8
WA8
WE7
WB7
C08
WA3
ED6
P18
H2A
C05
H3A
P05
WC4
EB5
P22
H2B
H6B
EB4
H5A
WD3
W13
WH4
EB2
WG3
WG6
WG5
WH3
WH8
W16
W18
WG7
WH6
EAS
EA8
WIS
WG4
EA4
C16
WF5
C15
C14
C12
C13
C01
P02
Pil
WC6
H3D
EC6
WB2
11/7/77
11/7/77
11/23/77
11/23/77
11/29/77
12/1/77
12/16/77
12/16/77
12/23/77
12/23/77
12/27/77
12/30/77
3/1/78
4/29/78
6/2/78
7/9/78
7/21/78
7/30/78
8/3/78
9/1/78
9/1/78
9/15/78
9/20/78
9/21/78
11/17/78
12/12/78
12/18/78
12/18/78
12/19/78
12/19/78
12/20/78
12/20/78
12/20/78
12/20/78
12/20/78
12/21/78
12/23/78
12/26/78
12/26/78
12/27/78
12/28/78
1/5/79
2/5/79
2/7/79
2/8/79
2/9/79
2/21/79
2/21/79
2/23/79
3/30/79
4/13/79
6/15/79
6/15/79
7/13/79
7/14/79
124
$36,750
$39,500
$45,500
$43,900
$41,900
$52,688
$64,492
$52,988
$52,688
$65,000
$47,575
$46,000
$58,500
$55,500
$61,000
$57,000
$50,000
$36,770
$32,500
$67,000
$55,000
$46,000
$48,000
$59,400
$44,700
$55,800
$44,500
$50,000
$44,500
$39,000
$45,000
$38,500
$65,800
$55,000
$79,100
$65,800
$45,000
$36,000
$55,000
$56,300
$38,500
$55,500
$67,900
$34,000
$66,900
$66,900
$66,900
$66,900
$66,900
$66,500
$68,000
$47,000
$64,000
$61,500
$46,000
Table A3.2
W14
WG1
Wil
WG2
WG8
WH2
W12
ED4
WH5
WH7
P14
W15
EA3
H1B
WE 1
WA3
W17
C09
P23
EA2
WH8
EA5
WHl
EAl
WD4
H2A
EA8
P1o
WD2
WD8
P23
WG3
Wil
ED7
C05
H5A
C12
C08
H2A
WE5
WB7
WC1
H3B
H3A
EB6
WE5
W13
EA3
C04
C15
C14
WG7
EB1
ED8
EC 1
8/10/79
8/13/79
8/17/79
8/31/79
8/31/79
8/31/79
8/31/79
9/6/79
9/7/79
9/17/79
9/19/79
9/22/79
9/28/79
10/24/79
11/1/79
11/16/79
11/16/79
11/29/79
12/15/79
12/21/79
1/4/80
1/25/80
7/11/80
7/18/80
9/5/80
9/19/80
11/3/80
11/21/80
12/29/80
12/30/80
4/3/81
10/9/81
10/29/81
10/29/81
11/14/81
12/4/81
12/16/81
12/31/81
1/5/82
1/7/82
2/4/82
3/29/82
1/4/83
1/10/83
2/28/83
5/15/83
6/17/83
9/30/83
9/30/83
10/30/83
11/16/83
11/18/83
11/25/83
11/28/83
12/20/83
125
$59,800
$47,500
$57,500
$41,000
$69,800
$47,500
$59,300
$60,500
$42,000
$69,800
$60,000
$63,500
$42,000
$60,000
$60,000
$62,000
$84,100
$75,000
$78,000
$73,000
$80,000
$45,000
$53,950
$50,950
$70,000
$67,200
$72,500
$65,000
$66,000
$97,000
$86,000
$73,000
$76,000
$52,500
$87,200
$78,750
$103,000
$88,000
$74,000
$69,200
$84,000
$45,000
$76,500
$78,500
$90,000
$77,000
$88,650
$58,050
$86,000
$114,500
$113,850
$103,500
$60,500
$93,000
$63,000
9z i
009'El 1 $
O00'0Z LS
009'90 I S
000,99S
006'LLS
000'90 L$
009'17eL
000't Is
00O'0I I S
OOS'L9$
L9/t/6
Z9/ t /Z
9/02/9
LZ/I
L/
LB/I 1 /9
L9/l/9
L9/9t/C
000'06 L
49/6/0
B'/OC/ZL L
99/0C/l
000'06$
000'ZL$
99/6Z/ZL
LOd
9a3
817H
1GM
9V3
89H
9tO
GM
ZOd
983
C3M
91M
000'99t S
09L'99S
000'9$ s
000'176$
O00'6LS
009'36$ I
000'tt$S
000'00 S
000'90I$
006'99$S
000'09$
000'99$
000'Z9 LS
009'U9$
000'96$
000'176$
000'VLS
000'Z6$
09Z'Z6$
000'M LS
99/ t t/l
99/96 /L
99/ L/9
1V9/9t/9
99/9/9
903
VEH
183
CHM
90d
91M
G7H
933
zIM
9td
93M
91m
LVM
09H
90d
t03
ZV3
9V3
ZG3
GIH
V133
COM
OPH
1G3
Z83
V9H
1703
VGM
9dM
600
93M
Old
95M
VIH
17CM
G0H
99/6/I
IVM
99/0/6
99/91/6
99/lL /6
99/9/9
99/9g/t
99/C t/96
99/91
99//LI
98/17/1
99/9t/OL
99/I/0Il
99/9/L
000'00 L
000' 1701
S 99/9t /
000'06$
99/9/
000'17LS
99/92/I
000'69S
t79/0It/El
000'eL$
006'90 L$
000'910$
009'L L$
000T9$
009'99$
09L'OQS
000'L LS
000'08$
000'99S
179/l/ll I
99/0Z/L I
179/ll/0l
Zftaq'ki
-
Table A3.2
wil
C14
C13
WB5
EA3
WD6
WE6
HIB
WG2
HA
EA6
EE8
P03
WG1
Cll
P02
EC6
EA8
H6D
WC1
WC4
WF2
WAl
WE8
C08
WG7
WG8
EC5
WH8
W17
W15
H5A
WB7
WB6
EB2
WF8
WE7
H2C
WH4
WEl
EAl
10/30/87
11/9/87
11/12/87
11/20/87
12/4/87
12/11/87
12/31/87
3/26/88
4/28/88
5/6/88
6/28/88
9/30/88
12/2/88
12/8/88
1/2/89
2/21/89
7/29/89
8/9/89
8/31/89
9/8/89
10/16/89
10/27/89
12/13/89
12/29/89
3/7/90
4/20/90
10/5/90
11/30/90
1/17/91
7/16/91
8/30/91
10/21/91
12/27/91
1/6/92
3/27/92
4/2/92
4/3/92
4/28/92
5/15/92
1/29/93
1/29/93
127
$88,000
$146,000
$145,000
$81,900
$73,000
$96,500
$73,500
$113,500
$59,900
$113,500
$120,000
$125,000
$115,000
$80,000
$140,000
$125,000
$115,000
$108,000
$115,000
$57,000
$83,200
$72,000
$84,500
$91,000
$113,000
$95,000
$89,000
$60,000
$85,000
$73,200
$62,500
$100,000
$86,000
$56,250
$78,500
$87,000
$88,000
$112,800
$52,000
$56,000
$47,000
Table A3.3
Club
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
P npmnnt
II~rdamont
I Init~ I Pc~ntIw~
r'nInnv r'I"h
Re I Dznf /wo I
IColonv
# Hnif-q
I
Ant /WP
Condominium Rents
Condo
Condominium Project
Rent
Adjusted
|Pico Townhouse |Whiffletree
|Hemlock Ridge
Value
I # Units | Rent/we |I # Units | Rent/we I # Unit Ien/we Index
nt/we
# nits I Rent/we # nits
130
124
80
175
175
200
200
238
288
313
325
350
388
475
413
475
540
468
600
672
750
670
115
123
133
143
143
158
168
180
237
313
388
346
400
400
406
368
368
526
320
284
175
175
175
175
200
238
263
288
335
380
395
435
460
480
580
660
622
622
600
160
188
213
263
300
318
455
475
475
140
150
160
163
180
193
210
237
310
380
348
490
400
406
430
490
420
410
316
80
115
149.433
154.5294
166.1176
169.9381
192.1856
218.7216
239.5567
268.7526
322.2353
392.3299
391.3711
449.2887
423.4706
443.2471
444.1882
507.2941
523.8588
486.7529
419.3647
128
133
Table A3.4
Hotel Rents
Hotel
I.
lInn (9 Lona irail
IChalet Kilingto~~ ~ortina Inn
IRed Rob Inn
Inn
Cortina
# Rooms Rate/do
Rate/do
#
Rooms
Rate/do
#
Rooms
# Rooms Rate/dI
-.
L-ear
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
55
55
55
55
55
55
55
55
55
55
55
55
55
55
55
55
55
55
55
55
55
55
55
36
39
39
42
46
50
49
52
60
70
80
92
96
100
120
125
125
123
150
160
160
141
130
57
57
57
57
57
57
57
57
57
57
57
57
57
57
57
57
57
57
57
57
57
57
57
38
41
42
45
58
60
56
64
64
70
75
84
97
110
90
75
86
98
116
120
120
120
120
.
90
90
90
90
90
90
90
90
90
90
90
90
90
90
90
90
90
90
90
90
90
90
.
30
32
30
32
33
35
41
45
54
60
65
70
75
88
89
115
152
180
140
151
148
150
-
32
32
32
32
32
32
32
32
32
32
32
32
32
32
32
32
32
32
32
32
32
32
32
40
39
40
40
39
43
47
48
56
67
69
88
98
108
126
134
144
153
158
158
158
168
168
Rent
Inde
37.68056
36.02564
37.17521
37.84188
42.5812
44.94017
45.04701
50.1453
54.65812
63.4359
69.58547
79.11966
86.51709
93.91453
101.2051
100.2051
114.2521
132.1667
154.3504
142.2906
146.5214
142.2692
140.453
Table A3.5
Sherburne Condominium Projects
5-29-90
Units
Name
Colony Club
34
Edgemont
40
Fall Line
36
Glazebrook
44
Hemlock Ridge
24
Highridge
119
Inn of Six Mountains
103
Killington Center
20
Moon Ridge
20
Mountain Green
216
Northbrook
10
Northside
12
Pico Townhouse
24
Pinnacle
'150
Sunrise
172
Telemark
31
Trail Creek
80
Trailside Village
11
Valley Park
16
Village Square at Pico
132
72
Whiffletree
Winterberry
4
Woods at Killington
107
1477
130
......
......
......
**.*
Bradford, Susan. "Second Homes", Builder (February 1991), pp. 79-90.
Carliner, Michael. "Second Homes", HousingEconomics (July 1990), pp. 8-10.
Davis, Jerry C. "Hospitality Firms Strive to Compete in Resort Market", NationalReal
Estate Investor (July 1990), pp. 50-60.
Dumaine, Brian. "Investing for a New Future", Fortune (January 27, 1992), pp. 68-71.
Employment andEarnings,States andAreas, Washington D.C.: U.S. Department of
Labor, Bureau of Labor Statistics, 1969-1992.
Hawks, John W. "The Next Boom in Real Estate", American Demographics(May 1991),
pp. 46-48.
Hawks, John W. and Nevins, Faith P. "Recreational Real Estate - Follow the Baby Boom
Into Resort Development", Development Magazine (November/December 1991), pp.
13-15.
"How Big is the Second Home Market", Builder (October 1990), p. 72 .
Jacobs, Harvey S. "The Baby-Boomer Boom", Regardies (April/May 1991), pp. 23-25.
Lachman, M. Leanne. "Leisure Time in the 1990's", Urban Land (August 1992), pp. 2228.
Land Use in Transition:EmergingForces & Issues Shaping the Real Estate
Environment, Washington D.C.: Urban Land Institute, 1993.
Logan, Gregg T. and Day, Ann E. "Ski Resort Real Estate Development", UrbanLand
(December 1989), pp. 36-37.
Miller, Norman G. "Residential Property Hedonic Pricing Models: A Review", Research
in Real Estate (1982, Volume 2), pp. 31-56.
Phillips, Patrick. "A Resort For All Seasons: Hidden Valley, Somerset, Pennsylvania",
UrbanLand (March 1987), pp. 20-23.
131
Phillips, Patrick. Developing with RecreationalAmenities: Golfy Tennis, Skiing and
Marinas,Washington, D.C.: Urban Land Institute, 1986
"Resorts and Recreation", UrbanLand Special Trends Issue (March 1990), pp. 32-3 5.
"Resorts and Recreation", UrbanLand Special Trends Issue (March 1991), pp. 37-39.
Saporito, William. "Recession Thaws At Ski Resorts", Fortune (January 27, 1992), p. 13.
Smart, Eric J., et al. RecreationalDevelopment Handbook, Washington, D.C.: Urban
Land Institute, 1986
Smith, Valene L. "Recreation Trends and Mountain Resort Development", Journalof
Travel Research (Fall 1991), p. 53.
Staples, Ed. "Second Homes May be Poised for Boom", Real Estate Finance Today (June
3, 1991), pp. 8, 11.
Suchman, Diane R. "Opportunities for Recreational Developers", UrbanLand (February
1991), pp. 32-33.
Symonds, William. "Ski Resorts That Go Beyond The Slopes", Business Week (January
16, 1989), pp. 106-107.
Terleckyj, Nestor E. and Coleman, Charles D. RegionalEconomic Growth in the United
States: Projectionsfor1991-2010, Washington D.C.: NPA Data Services, Vol. 1,
1991.
United States. President. Economic Report of the Presidenttransmittedto the Congress,
Washington D.C.: G.P.O., 1992.
Waldrop, Judith. "Who Owns Recreational Property?", American Demographics(May
1991), p. 49.
"What Do We Know About Second Homes?", Economic News Notes (May 1987), pp. 69.
132